Federico Imparatta
Driving Scalable Product Delivery, AI Integration & Global O
New to Contra
Federico is ready for their next project!
people talk about “ai in healthcare” as if it’s a slogan. the real work is getting models to behave in the real world, with messy data, uneven labeling, and edge cases that don’t care about your roadmap.
at ataraxy, i led the machine-learning stream for a large clinical data product. we rebuilt the entire pipeline: data normalization, feature extraction, model training, evaluation, and deployment. we dealt with inconsistent medical records, contradictory annotations, shifting taxonomies, and the usual parade of surprises you only find once you start training at scale.
we added explainability, tightened feedback loops with domain experts, and set up the automation so retraining didn’t depend on heroics. the result was a model that performed well, generalized beyond the sandbox, and actually shipped.
13
82
most data platforms rot from the inside out. too many pipelines, too many one-offs, too many decisions made under pressure and never revisited. everyone feels the pain, nobody has the time to fix it.
my role over the past months was to stop that cycle. i took ownership of a fragmented data environment, rebuilt the foundation on snowflake, and cut operating costs without slowing delivery.
we consolidated historical loads, redesigned the sdlc, killed redundant compute, cleaned up storage waste, and rewired ingestion so teams could trust the data again. i pushed for clear ownership, predictable backfills, and a migration plan that didn’t break downstream consumers.
the impact was simple: faster delivery, lower spend, fewer surprises. a data platform that finally behaves like a product instead of a patchwork of urgent fixes.
16
109
ataraxy: product strategy and end-to-end delivery
ataraxy began as a small product practice focused on turning rough ideas into working products. it grew into a team delivering full-stack experiences: software, mobile, backend services, integrations, connected hardware, and applied AI.
my role
i set product direction and built the operating model. the goal was simple: a team that could ship without relying on me.
what we shipped
• end-to-end products in multiple industries
• data pipelines, AI workflows, and connected experiences
• unclear client ideas turned into products with real users
• consistent delivery and technical clarity
• in 2024, white pearl technology group acquired 50% of the company
1
17
most ai projects talk about “trust” without ever earning it. at verusen we took the opposite path: we built explainability into the core of the product, not as an accessory.
my role was to make this real. i led the work that turned years of messy historical recommendations, user notes, risk flags, procurement logic, and domain heuristics into a system that can finally explain why the ai suggests what it suggests.
we mapped recommendation patterns, rebuilt the rules engine, audited edge cases, aligned data teams and product, and created the loop that lets users challenge, correct, and strengthen the model. no black boxes, no magic. just clarity.
the result is an explainability layer that lets mro, supply, and procurement teams interrogate decisions, understand tradeoffs, and act with confidence.
17
108