Hands-on experience in data products and infrastructure, fostering cross-functional collaboration across technical, product, and commercial teams. Focused on data-driven decision-making built upon a robust background in statistical analysis, predictive modeling, algorithm design, machine learning, and data/text mining.
Successful leadership of complex technical projects in both B2B and B2C environments, proficiently managing remote teams, and actively contributing to talent acquisition and mentorship.
Specialist in optimization problems, productization of ML, personas segmentation, multichannel strategies and driving conversion uplift. 17+ years of coding experience in data proc/viz, cloud/grid computing, web, data warehousing, ETL, front/back end.
Developer by nature on economical self sustainable projects. Driven by innovation, novel business models and collaborative working environments. Passionate about story telling and crafting, good knowledge of international languages and long coaching experience.
.
Years of
Data
Languages
Data analytics
95%
Data science
85%
Data platform
80%
Data governance
90%
Tech lead
95%
Prototyping
85%
Client facing tech presales to scope and qualify opportunities. Experienced in narrowing down the scope and design a solution attractive to the clients.
Hands on in creating insights from data and guide business data drive decisions.
Advising clients to select the best tech stack and governance model to enable better operativity.
Hands on in creating prototypes to showcase clients in few days how a solution would work in their case.
Softserve | 02-23/present
Tech lead for scoping and implementing AI projects. Building and steering the teams responsible for implementation. Positioned between hardware producers (nvidia, lenovo, dell, cisco...) and the market to consult, teach and realize ML/AI projects with the use of the most modern solutions both on the appropriate hardware (cloud/on-prem) or software.
startups + lightmeter@Y-combinator | 09-22/01-23
Consulting early stage start-ups to architect and deliver data infrastructure: storage, interfaces, middlewares, scheduler, messaging, security, redundancy, analytics, BI. Train and implement language models with focus on email outreach.
afiniti | 09-20/07-22
Leading the technical implementation of afinti AI solutions for big European clients (telco, banking, media) and managing remote teams assigned to the project: DE, DA, DS, AI, DB admin. Responsible in customer projects: pre-sales, feasibility study, metric design, data reliability, data analytics, performance monitoring, billing, financial reports.
circ/bird + vay | 09-19/09-20
Developing a full stack (micro service based) solution for fleet dispatch. The engine defines the most profitable task assignments for the capable agents on the field. Demand forecast and customer patterns. Forecast on telemetry data.
Motionlogic@Deutsche Telekom | 10-17/09-19
Interpreting mobile data of 30M users and their mean of transportation, product owner for product development of origin-destination matrices, responsible for customer deliveries (mobility, advertising, real estate, commuter patterns). Machine learning, insight highlights and data monetization.
Mediamond@Mediaset | 10-16/10-17
Data driven advertising via audience segmentation on a large cross device media network. Responsible for target quality and performance, inventory forecast, modeling, data visualization, and business intelligence. Tech consultant for user profiling built from web logs, CRM data, second party enrichment, semantic engines, and SEO keywording. Representing my company for the big data project of the corporation.
E-dynamics@Lufthansa | 10-14/01-16
Data driven segmentation of price and ancillary teasers on lufthansa.com. Campaign design, tracking, revenue calculation, reporting and tool concept development. Owner of 30+ A/B tests with 3M certified revenue uplift.
dauvi | 02-13/10-14
Advising SMEs to organize and collect data in their companies, installing cloud based ERP/CRM systems. Setting a geo based rural portal to connect professionals.
UniGoe | 06-09/01-13
Developing a parallel high performance c++ suite for Monte Carlo and Molecular Dynamics simulations, openGL 3D visualization, ETL, data visualization and reporting for biological oriented questions. Publications: Influenza Fusion, String Method, Pore Formation.
DE: automotive | 3m | skelarn, PLC, scada
Data driven solutions can improve the efficiency of factories, especially regarding ventilation and heating. Here with the help of data we proposed 10 actions which can bring 10%-20% savings in energy consumption.
Conduct the client visit, visit the factory and the data room. Talk with all stakeholders and design an action plan for energy efficiency
KPIs/Impact: 10%-20% energy cost improvement
DE: automotive | 1m | d3, llm, python
Managing a fleet requires a stable infrastructure and a prompt resolution of issues. End clients sometimes have issues in using the portal and first level support can be simplified by agents.
Conduct the client visit, design the implementation plan, create the mockup for the client to test
KPIs/Impact: increase response time, error resolution and customer satisfaction
DE: energy | 3m | python, skleran
Managing a large pipeline infrastructure need 24/7 human support to be able to act in case some problem arise
Collect all the data for the Bavarian gas subnet, create the graph, map the quantities to the respective nodes, run simulations, bechmark ML results
KPIs/Impact: Increase accuracy, system stability and predictive maintenance
DE: public | 1y | cloud and on-prem computing
Government offices and public utilities need important modernization on their digital environment. Processes are not traceable and interfaces lack capabilities to connect with modern tools like agents.
Collect all the requirements and prepare the specific questions regarding the client tech stack and needs. Prepare an implementation plan following the guide line of the RFP and explain how the execution would be enrolled for the length of the project, usually across multiple year.
KPIs/Impact: Precision of solution and requirements for better RFP acceptance
DE: pharma | 3m | snowflake, databricks
A complex manifacturing corporate need a strategy to design a data platform to enable agents to support their workforce. We consult the client in designing a platform that enable the integration and cosinstency of data sources and the implementation of agents
Conduct the visits with all business units, collect requirements and discuss feasible solutions, propose a 2y migration plan
KPIs/Impact: Submit the data platform architecture to enable proper data modeling and agentic workflow
DE: automotive | 5m | nvidia morpheus
Clients struggle to parse security critical logs and they need GPUs to effectively run dedicated ML models to anticipate threads and spare batch processing. The added efficiency improves operations and reduces storage and batch processing. A model should be trained on their own workflow and specific data.
Collect the requirements, sync with nvidia about the implementation plan across 200 locations and 200k employees
KPIs/Impact: train a model to run on nvidia GPUs for the specific customer need
DE: pharma | 3m | snowflake, databricks
A complex manifacturing corporate need a strategy to design a data platform to enable agents to support their workforce. We consult the client in designing a platform that enable the integration and cosinstency of data sources and the implementation of agents
Conduct the visits with all business units, collect requirements and discuss feasible solutions, propose a 2y migration plan
KPIs/Impact: Submit the data platform architecture to enable proper data modeling and agentic workflow
DE: automotive | 5m | databricks
The client is too slow to fix some issues in their fleet and they need tools to predict the next issue to promptly plan maintenance. The platform should correctly collect and parse telemetry data and ML should reliably predict issues.
Conduct the client visit, talk with the relative stakeholders, run exploratory analytics and prepare the implementation plan
KPIs/Impact: present and implementation plan based on an accurate data analysis
DE: insurance | 3m | llm, python
The insurance contract is pretty complex and the client need quick support to help their employees to respond to client's questions.
Devlop the tools to parse the contract information, tweak language specific issues and run tests and benchmarks to find the most suited model
KPIs/Impact: A functional MVP for their insurance consultancy bot
DE: bio | 6m | llm, python
The client has highly specialized scientific products which require highly qualify tech sales assistance which is hard to find. A well tuned RAG supports the client in finding a quick and reliable solution
Parse all the documentation and develop the best suited tools for better accuracy. Review the infrastructure design and propose the selection of better suited tools.
KPIs/Impact: Overview the infrastructural issues preventing the agentic system to deliver the expected quality
US: hardware | 4m | -
Advising hardware resellers how to scope a business opportunity to propose their clients further investment in hardware facilities.
Preparing the presentation material, travel in different countries, talking in different languages to the local salesforce.
KPIs/Impact: leads and opportunities
EU: tech | 9m | LLM, python, fastapi, D3.js, nodejs, aws, docker
Project Description: Complex and long information can be challenging to understand and remember without mind maps that represent the most important concepts. Booklink summarizes the most relevant concepts and the connection between them displaying in an interactive graph. Language models summarize the content of the text and the connections between the nodes.
Deciding the tech stack productization of the code interacting with the language model Increase code quality for front-end, back-end and batch processes ML ops on aws and on prem
KPIs/Impact: deliver the MVP
DE: tech | 6m | Airflow, metabase, kafka, red-panda, traefik, nginx, postgres, python, go
Emailing in the 20s is still the most relevant mean for sales conversion but email delivery gets more and more challenging as providers improve their filters by recently adding language models. The company needed an analytics platform to understand the main KPIs of their business and operational efficiency.
Collect requirements Propose the most suited data platform for fulfilling company's needs Implement the data platform on-prem with docker. Develop two custom middlewares Expose and document custom APIs Introduce tests and health checks
KPIs/Impact: deliver the complete data platform
DE: tech | 4m | Nodejs, angular, LLM, custom middleware
Sales emails need to be effective and concise, language models can help supporting sales force to make their messages more effective.
Ideate the solution and pick the tech stack Discuss technology moving components and change solutions as market tools change Fine tuning vs prompt eng vs embeddings Custom test models Write the prompts and the code Deploy the MVP
KPIs/Impact: implementation strategy for startups
UK: telco | 1y | R, sql, stan
Project Description: Sales agents motivation is crucial to our client success and the best delivery for our AI solution. Compensation should be aligned to the most important client KPIs and the optimization metric. Which scenarios are the most effective for delivery the highest gain simulating from real life data?
Define with the customer the most important KPIs Collect the data and design the simulation Develop the models corresponding to each metric Simulate scenarios where each agent compete with other agents on each metric Assess and present the best strategy as return of investment
KPIs/Impact: Increase CLV by aligning company goals with agent compensation
IT: bank | 6m | Postgres, sql, python
Project Description: Mortgage installments can be paused for different reason but the credit institute needs to restart collection incentivizing the debtor to find an agreement. An AI solution support the uplift in conversion for resuming the installment collection.
Dig into data sources, build KPIs and propose a metric Iterate with the AI teams to estimate an uplift on the metric provided by the ML model Iterate with the DE teams to build the pipeline to feed data into the AI model Test metric biases and propose the client operational changes to increase optimization capabilities
KPIs/Impact: Increase mortgage installment payment
UK: media | 1y5 | Postgresql, stan, sklearn, R, gcp
The company offers an AI solution for optimization. The metric has to be discussed with the client for providing the most effective business and operational value while being measurable, consistent and built on reliable data sources. The metric performances should be stable over time and its uplift forecasted to prevent mis expectation with the customer and internal stakeholders.
Sit with the business/operational client teams and understand their needs. Sit with the tech client teams and understand data sources and connectors. Propose a metric that is measurable, consistent and aligned to what our AI model can deliver. Customer lifetime value metric calculation. Predict an uplift and build the commercial proposal Instruct the DE, DA, DB, AI teams on how to work with that metric Test and monitor the processes, keep the client in the loop for reporting,
KPIs/Impact: Build reliable metrics and pipelines to reduce inconsistencies with sources and client reports
IT: telco | 1y | Aws, gcp, Greenplum, IBM, oracle
The client wants to move from their on-prem data infrastructure to a big cloud provider.
Sit with the client teams and discuss all data sources and integrations in use. Propose which processes can be shifted on the cloud making sure efficiency, security and costs improve Work on every single pipeline checking data integrity and reporting.
KPIs/Impact: submit an on-prem data platform
DE: telco | 1y | Sql, python
The client wants to implement our company AI solution but it is not sure whether the technical implementation is feasible nor the revenue goals can be reached.
Understand customer needs and tech stack Propose a technical solution and a commercial proposal. Consistency check on data quality and business estimates Review contract.
KPIs/Impact: revenue forecasts
DE : mobility | 3m | Tensorflow, keras, ML, python, spark, kinesis
Remote driving needs a reliable technology to enable drivers to log into connected cars and drive safely. Internet connection shows issues in providing reliable connection with cameras and car controls. A ML investigation need to find the root causes for disconnection
Collect all useful data and assess data quality Feature engineering and selection. Model selection and hyperparameter tuning Debugging of data inputs and re-phasing of information. Spot problem and feedback to engineering team.
KPIs/Impact: reduce downtime
MY: pharma | 5m | Aws, python, angular, blockchain
Counterfeiting is a serious problem for pharma. Middleman warehouses in some countries swap packages and replace the original product with cheaper medicaments. Blockchain can ensure traceability and spot the transaction where counterfeit could have token place.
Decide the tech stack. Analyze existing data and estimate the phenomenon. Build app MVP Talk to investors and partners
KPIs/Impact: reduce counterfeits
DE : mobility | 6m | Aws, celery, redis, sklearn, tensorflow, python, fastapi
Optimizing workforce in a vast operational area is crucial for company profitability. Workforce needs to drive from warehouses to each scooter in a city for operational tasks. A model is required to distribute workforce considering multiple constrains (working hours, travelled distance, tasks number…) that accepts multiple status changes and keeps optimizing task distribution without disrupting current operations.
Go on field and collect all the task types and problems connected with operations Build graphs and routing software and put them in production Build a model to create and iterate routes and evaluate their efficiency Generate myriads of simulations and tune parameters. Deploy the model and test the API connection. Develop a middleware to convert the different data types Deploy the services on difference VPCs, configure load balancer, test Compare solutions with other software and present it
KPIs/Impact: improve ops efficiency
DE : retail | 2y | Sklearn, tensorflow, python, spark, yarn
The client needs to understand the potential of their user base comparing car drivers stopping at their stores compared to the ones stopping at competitor stores. Mobile network data is analyzed to understand motorway traffic and stops during users journey.
Build filter to subset the user base to match customer requirements Iterate with the engineering team to improve software quality especially regarding routing Send the computation and transform the data. Analyze the data and compare with external sources. Build crawlers to retrieve updated information from internet to tune parameters. Use machine learning to further improve matching with external data and extrapolation factors. Build recurrent reports and keep the conversation with the client including future feature development
KPIs/Impact: increase conversion rate for retail
JP : automotive | 2y | Hadoop, spark, yarn, python
A big electric car manufacturer needs to know the mean travel distance in German motorways to plan the construction of electric charging stations for the given capacity of their cars. Mobile network data supports the estimation on how many users drive daily on the motorway network and sophisticated routing algorithms compute the exact junction points crossed.
Iterate with the engineering team about the quality of the routing solution, testing and re-process data for each iteration. Develop spark functions to process the parquet files and extract the data needed. Build reports for the client and add consistency checks for the discussion. Data visualization and presentation at customer location
KPIs/Impact: planning of electrical charging stations
EU : retail and real estate | 6m | Hadoop, spark, yarn, python
Retail stores need to understand the number of people passing in front of the building to establish the potential, the capture rate, the sales rate and building value. Additional information can be added to this user pool as origin, nationality, expense capability and socio-demo from their mobile data traces.
Define all the polygons covering the stores Define all the mobile cells covering the polygons. Develop a software to re-center user position based on cells information. Subset the user dwelling or passing by the polygon. Build reports and present them.
KPIs/Impact: Real estate properties value
US : restaurants | 1y | Tibco, qgis, python
Retailer has a lot of data about sales compared to the geographical position of the store and they want to know what other locations can be suitable to open a new store with similar revenue benefits.
Analyze customer data, calcolate isochrones. Enrich customer data with dwellers report to calculate capture rate Analyze country wide data to spot locations with similar conditions outside of the isochrone areas covered by existing locations. Build report, data visualization and present to customer.
KPIs/Impact: Planning for new retail locations
SE: telco | 2y | Hadoop, yarn, spark, python
A telco company operating in other countries established a partnership with us to enter their market with products similar to the ones we sell in Germany and they need consultancy to build the software to analyze their data.
Identify the most important stakeholders and establish the partnership. Prepare presentations and sessions for the knowledge exchange. Work on their reports and point out inconsistency and improvements. Support with their commercial proposals
KPIs/Impact: Data monetization
DE: mobility | 6m | Hadoop, yarn, spark, python
A transportation company needs to know how many users use their services: where they start their journey, where they enter and leave their means of transport.
Iterate over different filters to subset the user base into daily commuters only Iterate over different algorithms to perform modal split Use the limited external data to tune own model parameters. Run the calculation and build the reports for the customer.
KPIs/Impact: Improve public transportation usage
IT: media | 2y | Tealium, pubmatic, bluekai, krux, optimizely, nugg.ad
In advertisement real time bidding is a technology that allows multiple players to trade creativities over internet. On top of seller and buyer platforms other platforms (data management platform, header bidder, ad server…) need to be synced to participate to the bid and conclude the trade within 50ms. All of these platforms need to be configured and synced via scripts or tracking pixels.
Speak with the vendor, test and configure different platforms: SSP, DSP, DMP, ad server, tag manager, optimization tools, header bidder. Speak with the service provider to prepare custom services for information syncing Dump the enriched data to the datalake
KPIs/Impact: Increase targeted spending
IT: media | 2y | Aws, gcp, azure, python
Multiple companies share the same revenue channel which is the sales house I represent. Data need to be shared into less platform to enable monetization. External platforms should dump data into a datalake to further enrich user profiling and useful targets for the industry.
Talk with cloud providers (aws, gcp, azure) evaluate their proposal and test the solutions Define all the platforms that need to be integrated. Convince the stakeholders to participate to the program and support them in the migration. Scout technology providers that can support the coordination of data sources and pipeline ingestion
KPIs/Impact: Datalake foundation
IT: media | 2y | Tealium, pubmatic, bluekai, krux, optimizely, nugg.ad
Targeted advertising has much better conversion rates than any random advertisement. Good targets are hard to find and perform differently under different metrics. Data enrichment is a key component and first party data need to be enriched by external providers whose quality and reliability has to be proven.
Define the valuable target for the industry, build a catalogue of audience Expand the targets across the user base using look-alike. Post evaluate the data especially for socio-demo. Assess target performances and use the data to improve target definition Enrich the data with third parties: retail and telco data. Scout the provider with the best linguistic model to label content and import the labels. Sync the data with relevant platforms. Monetize the data
KPIs/Impact: Targeted audiences for advertising
IT: media | 2y | R, python
Sales force needs to sell two weeks in advance the available inventory. Accurate forecasts are essential to fulfill all available spots.
Define all the different inventory types. Implement different forecast methods and tune them. Run a loop on historical data and watch the competition among models per inventory type. Select the most performant model per inventory type. Run the forecast each week and communicate with the marketing team the available inventory for the next two weeks. Analyze the sold and unsold and iterate for the subsequent week
KPIs/Impact: Improve selling inventory
IT: media | 5y | sql, postgres, mongo, neo4j, elastic
Data insights need to be access quickly and many visualization types should be used to represent the most important feature of each analysis. Over years the environment of available tools changed a lot and solutions adapted to current trends.
Understand which data should be used by which stakeholder. Agree on the time aggregation and graph type. Build views and aggregated tables in the database Schedule the jobs for aggregated tables. Find the most suited BI Grant permissions or connect with other authentication tools
KPIs/Impact: data analytics tools for finance
DE: airline | 1y5 | Tealeaf, webtrack, maxymizer, R
The booking funnel is leaky because some users don't enjoy/understand the user interface and leave for looking at a competitor solution. UI/UX and personalization tests help the developer team decide for the best solution to put in production. A/B test is a reliable way of understanding which solution is increasing the overall sales of the e-commerce portal.
Identify the weak spots with the help of analytics tools. Design and propose a test to put in production. Convince the stakeholders presenting the feasibility study and revenue estimate. Talk to agencies and instruct them how to implement the test. Put the test live and monitor performances. Analyze the different variants and present the results. Calculate the revenue uplift and get confirmation from the controlling team about the correct attribution to the team. Iterate over 30 times in a year
KPIs/Impact: 7 digits more booking revenue
DE: airline | 1y5 | Tealeaf, webtrack, NLP, R, voice of customers
An important tool to understand the goodness of your portal is the direct voice of users and customers. The website has a chatbot, a feedback tool and interaction logging and customers write to company CRM. Users write on social and affect brand reputation. All of this text need to be parsed and analyzed to understand how successful the brand and the product are.
Collect the onsite data sources Collect the providers data sources. Crowl information from internet. Parse the data (stem, stopwords, clean). Build own specific dictionary. Run sentiment analysis. Add external tracking information. Gain insight from outstanding comments. Write report and let it circulate
KPIs/Impact: improve customer satisfaction
DE: airline | 1y5 | Tealeaf, webtrack, NLP, R, voice of customers
Different companies within the corporate group need to define a common strategy to organize customer information and build personalized offers and experiences. The information is spread across multiple silos and contact channels. The group needs to build a datalake and a strategy to contact all users through the different contact channels (e-commerce, email, call center, advertising, premium program) with a consistent communication and ad-hoc offering.
Define requirements for the datalake and analytical platform. Scout the different data sources and speak with integration providers on how to ingest the data Design the products that will access user information and how those tools will delivery content. Design the type of personalization for each communication channel Iterate with many consultancy agencies for every specific project Present and report project results to multiple stakeholders. Write report and let it circulate
KPIs/Impact: unify corporate data
IT & TZ: SME | 1y5 | python, java, sql, php, android
Different SMEs need to move from on-site IT to cloud solutions but they need an ERP flexible enough to fulfill their specific business needs.
Talk to client and understand their needs. Scout the most suited solution for their needs. Develop modules for specific business needs. Configure the software. Build ad-hoc middlewares
KPIs/Impact: custom ERP modules for SME
DE: University/biophysics | 3y5 | C, c++, Qt, OpenGL, openMP, MPI
Cells are made of lipid membrane that isolate the interior of the cell from the exterior but show many different behaviors that allow the cell to perform certain types of exchanges with the external environment. One of the most important is fusion where the roles of lipids and peptides is not well modeled yet.
Model the components of the simulation Write the code and define each iteration step Parallelize the code with map reduce Tune the code Run the code in the cluster and analyze the data Develop the data visualization suite for 3D data exploration
KPIs/Impact: provide the pharma industry with reliable insights about transmembrane stability
DE: university | 3y5 | c, c++
Part of a graduate school and a national research project to contribute to specific topics in biophysics and neuroscience. National and external investors were requesting progress reports and talks to track progresses in the funded research topics.
Teach and tutor students. Present to conferences. Present to internal progress reports. Prepare application for funds Publish articles
KPIs/Impact: deliver a cross collaborative research project
Native: Italian. Fluent: English, German, Spanish Intermediate: French, Portuguese
python, js, c++, c, R, spark, go (viz) openGL, Qt, GTK+
tensorflow, keras, sklearn, scipy, caret, pytorch
gan, reinforcement learning, forecasts, predictions, classifications
html5, css3, js, d3, react, angular
fastapi, celery, redis, nginx, traefik, nodejs, php
SQL, postgresql, cassandra, mongodb, neo4j, elastic (fs) hdfs, s3
git, gitlab, svn, CI/CD, pytest, docker
docker swarm, kubernetes, airflow, kafka, presto, terraform
kibana, grafana, metabase, powerBI, tableau
mqtt, mosquitto, arduino, esp32, teensy, seeed, reaspberry pi, jetson
computer vision, 3d modeling, reinforcement learning, embedded programming
numpy, gsl, CGAL, matlab, octave, maple, root (grid) MPI, openMP
(GIS) qgis (viz) mayavi, povray (CAD/3D) rhino, blender, cura
RAG, graphRAG, lanchain, langsmith, llamaindex
s3, iam, ec2, sagemaker, bedrock, ecs, cloudfront, lambda, athena
bigquery, kubernetes, pubsub, gemini
factory, kusto, ml studio, synapse, databricks, cosmos
cuda, ominverse, morpheus, nemo, nim, tensorrt, triton
UCS, meraki, splunk
Create an internal assistant to guide the workforce to solve specific problems
Internal assistant
RAGCreate sales assistants who can answer detailed questions on product features.
Sales
RAGCreating an enviroment to allow developers to write safe code implemeing the company specific runbooks and avoid competence leak.
Coding assistant
on-prem© intertino 2026. Designed by HTML Codex