What spiega is about
- Knowledge sharing
- Portfolio
spiega tech
shared documentation across projects
spiega tech
Spiega is a collection of articles/post written during 17y of career spanning over 3k3 source code files and 2k images. The main areas are around data science and the impact of using machine learning in operations and business. Some content is presented as slides or as automated generated summary in the overview page. For a more business oriented description of those activities refer to spiega business.
Figure 1: module library
During this time I tried to make the organize the diverse topics into some main areas which are:
management and client projects management
Here is a collection of experiences regarding project scoping, execution and reporting
- governance
- manage and design data platforms
- team management
- how to manage and kick off a team
- project tools
- what makes the execution of a project efficient
Figure 2: module library
consultancy and service provider consultancy
- dauvi
- ERP tools and firm consultancy, mainly focused on setting up and implementing ERP tools and connecting different company components
- intertino
- web services. Offer segmentation strategies, agent compensation models, lagged metrics, customer lifetime value calculation, contact channels main and control metrics, scooter movement analysis, auto sensors assessment, shared bike usage evaluation, and weather prediction models.
Figure 3: module library
optimization experience data_science
Collection of projects for optimization challenges
- geomadi
- articles on geo features, routing applications, location intelligence, motorway stoppers analysis, routing comparisons, and sensor triangulation.
- antani
- optimization engine (reinforcement learning, simulations).
- mallink
- simulation engine in python.
- allink
- C/C++ simulation code for biophysical membranes
Figure 4: module library
data science competences data_science
Collection of libraries used as support to data projects
- kotoba
- generative ML. ML painting, text generation, music composition, agent naming environment for redundancy in images.
- albio
- time series analysis.
- lernia
- machine learning library used across multiple projects. Regressions without neural networks
- deep lernia
- machine learning library used across multiple projects. Neural network models
- ndoe
- Equations of motion, ride behavior, mobility concepts, POI capture rate, activation potential, data quality assessment, telemetry data types, prediction anomalies, and spatial analysis.
Figure 5: data science competences
programming and devops dev
Collection of work to develops products and manage data platforms
- sawmill
- big data processing. Data platform basics, data storage applications, compliance with access rules for sensitive data, modeling concepts, security practices, webserver and networking principles, messaging systems, middleware to protect requests, cloud provider impressions, CI/CD processes, testing methods, scheduling jobs, log processing, UI and data visualization techniques, and programming documentation.
- intertino
- web services. Offer segmentation strategies, agent compensation models, lagged metrics, customer lifetime value calculation, contact channels main and control metrics, scooter movement analysis, auto sensors assessment, shared bike usage evaluation, and weather prediction models.
- malastro
- useful scripts to manage a laptop, a server, embedded devices, media… Mainly using bash, ffmpeg…
Figure 6: data pipe example
art and science science
Non commercial projects around art, music and science
- kotoba
- generative ML. ML painting, text generation, music composition, agent naming environment for redundancy in images.
- science
- scientific contributions. PhD defense/paper on membrane inclusion objects, master defense/thesis on nanoparticle stability, bachelor defense/thesis on ion diffusion in lattice, Fokker-Planck equations, Fluttuazione theorem, physics papers, computer science lectures, PLOS One paper on influenza fusion, PR Letters paper on string method, and Sciencedirect paper on pore formation.
- algorithmics
- computer science features mainly used in scientific computation
- viudi
- music tech including theoretical discovery of composition, electronics.
Figure 7: module library
hardware dev
- sciame
- IoT applications. Coil for pickups/sustainers, synth collection, DSP, effects, mechanical amplifier, no-keyboard keyboard (touch surface), MIDI router player, audio mixer with op amps, piezo buffer for impedance matching and noise reduction, and creative coding techniques including openCV, openFrameworks, and processing.
Figure 8: sciame library
management
This section collects the relevant experiences in project and team management. The whole list of the 40 projects is in portfolio.
Figure 9: management experience
project tools management
Being efficient in running projects is essential and (especially in an agentic world) we need to prepare a good background for those agents to operate
- agent workflow
- org files
- productivity with org files
- agent context
- instruct agents how to behave and be productive
- agentic interview
- an interview run by agents
- documentation present
- knowledge graph
- multimedia utils
Figure 10: module library
governance management
Data platform management
- 26-2?] LLM governance
- how to manage agentic systems
- 26-2?] storage governance
- avoid resource waste in storing and processing data
- 13-2?] management
- management experience
- 26-2?] data_sovereignty
- on-prem infra
- 26-2?] knowledge_graph
- auto generated locally
Figure 11: module library
team management management
Team management
- 13-2?] team composition
- how to set up a team, SoW
- 13-2?] team productivity
- what makes a team productive
- 13-2?] well being
- body health for productivity
Figure 12: module library
optimization
Libraries mainly focused on optimization engine (operation, mobility…)
Figure 13: module library
geomadi data_science
Geomadi was used to perform complex operations on a large network. The mongo database contained all the information regarding maps and POIs in Germany and we were analyzing spatial data regarding specific subset of a complex network. The library uses spatial queries and geometrical transformations.
- 17-19] graph creation
- for routing applications
- 17-19] location intelligence
- to enrich predictions
- 17-19] motorway stoppers
- analysis on motorway traffic and capture rate
- 17-19] routing comparison
- comparing routing solutions
- 17-19] triangulation of sensor data
- locate mobile users
- 17-20] spatial utils
- for GIS
Figure 14: geomadi library
antani data_science
Antani is an optimization engine for logistics building efficient routes inspired by ants. The routes grow as polymers and variate to maximize opportunity/cost. This is a mixture of MonteCarlo as built in my PhD thesis and reinforcement learning. Each move has different polymer expansion options and the reinLearn decides what action to take to improve overall optimization time.
- 19-20] antani concept overview
- 19-20] antani infrastructure design
- 19-20] antani microservice integration
- 19-20] antani kpi comparison
- 19-20] antani overview page
- 19-20] antani compared with routific
Figure 15: antani library
mallink data_science
Mallink is the python adaptation of allink which was writte 2008-2012 in C++ and used as engine for antani
- 19-20] mallink optimization engine
Figure 16: mallink library
ndoe data_science
Ndoe analyzes motion behaviors and builds mobility reports
- 17-20] equations of motion
- 17-20] ride behaviour
- 17-20] mobility concepts
- 17-19] capture rate for restaurants
- 17-20] activation potential
- 20-21] quality on telemetry
- 20-21] data types for telemetry
- 20-21] feature in telemetry
- 20-21] forecast anomalies
- 20-21] prediction on telemetry
- 20-21] spatial
Figure 17: ndoe motion patterns
data science
Machine learning and statistics used in different projects
Figure 18: module library
lernia data_science
Lernia is a library used across project to use ML. It divides into lernia and deep_lernia where the second one is built on keras
- 15-20] lernia feature selection
- 15-20] lernia library overview
- 15-20] lernia blind test on series forecast
- 17-19] convNet on time series
- 17-19] sensor filtering
- 17-19] skewness in predictions
- neural networks">overview
- machine_learning">models
- AI in industry">2025 update
Figure 19: lernia library
albio data_science
Albio is a library to work with time series. During my career I used a lot of series decomposition and forecasts.
- 15-20] albio forecast
- with exogen variables
- 15-20] time series
- forecast in production
Figure 20: albio library
deep lernia data_science
Neuronal networks
Figure 21: deep lernia library
programming and devops
Knowledge used across the projects. How we take big logs and process them to get the most important information out of the big volume.
Figure 22: sawmill library
sawmill dev
Big data, data engineering, data pipelines, machine learning ops
- 13-2?] data platform basics
- example designs and principles
- 13-2?] data storage
- applications and principles
- 14-2?] data compliance
- access, sensitive data, anonymization, retention
- 15-2?] data modeling
- concepts and applications
- 15-2?] security
- practices and examples
- 12-2?] webserver and networking
- for websites and traffic to containers
- 20-2?] messaging system
- for stream and IoT
- 17-2?] middleware
- to protect and speed up external requests
- 15-2?] cloud providers
- impressions and experiences
- 15-2?] CI/CD
- praxis and options
- 06-2?] test and checks
- types and necessity
- 17-2?] scheduling jobs
- monitor and plan batch processes
- 15-2?] log processing
- from web, Iot, applications
- 06-2?] UI and data visualization
- for the business and client’s insights
- 12-2?] admin tools
- collection of script to admin a system
malastro dev
Development, sysadmin, prototyping
- 06-2?] programming praxis
- documentation, conventions, processes
- 02-2?] emacs
- mighty editor
- 06-13] c++
- simulations, native visual applications, openGL
- 13-2?] python
- data exploration, visualization, ETL
- 05-17] R
- data exploration, visualization, ETL
- 22-2?] go
- middelware to interact with relational databases
- 06-2?] SQL
- manage databases, create tables, join, format
- 13-2?] app
- native, angular, react, nodejs, cordova
- 06-2?] websites
- html5, css, bootstrap, php
- 06-2?] javascript
- openlayer, d3.js, maps
- 06-2?] node
- bot, automation, server
- 17-20] spark
- batch processing of logs
Figure 23: malastro library
assessments data_science
Coding interviews
- 19] scooter movement
- a portal for visualizing geo data
- 20] auto sensors
- analyze connected cars data
- 19] shared bike usage
- user usage of ride hail
- 17] ufo sightnings
- geospatial information
- 2?] weather prediction
- basics around data science
Figure 24: assessment library
consultancy and service provider
How to advice companies and clients to develop a digital/data strategy.
Figure 25: management experience
intertino consultancy
Web marketing and digital services, customer retention and communication strategies
- 13-17] offer segmentation
- 20-23] agent compensation
- 20-23] lagged metrics
- 20-23] customer lifetime value
- contact channels">principal and control metrics
Figure 26: intertino library
dauvi consultancy
SME consultancy, ERP systems, digital strategies
- dauvi
- presentation of the projects
Figure 27: geomadi library
art and science
Non commercial projects for art and science
Figure 28: art and science
kotoba art
Language and perception
- 19-2?] painting with ML
- generative ai
- 17-2?] text generation
- 04-2?] music composition
- 23-2?] agent naming environment
- 11-12] redundancy in natural images
- 99-2?] phonosymbolic language
- can we create a phonosymbolic language?
Figure 29: kotoba library
viudi art
Music development
- 13] entropy in music
- how to recognize the author of a song
- 23] music chart valuation
- how people judge the quality of songs
- 06] music composition
- generative melody generator
- 10] video production
- create songs with musical gears
- 10] video presentation
- description of the channel
- 10] video detail
- video detail
Figure 30: viudi library
allink dev
Software created during the PhD thesis
- 08-12] documentation
- for the project allink in C++
Figure 31: allink library
science science
Documents during the physics research period
- 09-13] PhD defense
- membrane inclusion objects
- 09-13] PhD thesis
- membrane inclusion objects
- 08-09] master defense
- nanoparticle stability
- 08-09] master thesis
- nanoparticle stability
- 05-06] bachelor defense
- ion diffusion in lattice
- 05-06] bachelor thesis
- ion diffusion in lattice
- 11-11] Fokker-Plank
- and master equations
- 07-07] Teorema della fluttuazione
- traiettorie a entropia negativa
- 02-09] Fizikaj Eroj
- klassika kaj kvuantuma mekaniko
- 13-13] Lezioni di informatica
- per piccole imprese
- 12-12] Paper
- Influenza Fusion, String method, Pore formation
Figure 32: fizikaj eroj library
algorithmics dev
Development of algorithms for better performances
- 11-12] configurational bias
- 06-12] Monte Carlo and molecular dynamics
Figure 33: algo library
media art
Collection of media projects
- 13-15] creative coding
- openCV, openFrameworks, processing
- 26-2?] blender animation
- language models and 3d modeling
- 15-2?] youtube channel
- manage a channel#+RESULTS: video-summary
- 15-2?] viudi channel
- current channel
- 10-2?] video production
- create songs with musical gears
Figure 34: media library
hardware
Sciame is a collection of projects for embedded and IoT devices
Figure 35: management experience
sciame dev
- 18-2?] coiler
- coiler for pickups and sustainers
- 21-2?] synth
- collection of synths for cheap microdevices
- 21-2?] dsp
- dsp and effects
- 15-17] McAmp
- mechanical amplifier
- 15-17] no-key-board
- keyboard with no keys (touch surface
- 22-2?] midi-hub
- midi router
- 97-2?] audio mixer
- with op amps
- 19-2?] piezo buffer
- for impedance match and noise cleaning
Figure 36: sciame library
script dev
cd $HOME/lav/src/spiegabash ./script/convert.sh
ls ~/lav/src/spiega/markdown/agent*