Hello, I'm

Kaustubh Lall

Senior Machine Learning Engineer · GenAI & Agent Systems

Production GenAI · Agents · RAG · Evaluation · Healthcare proof

I build production GenAI systems: conversational agents, RAG platforms, and the evaluation infrastructure behind them. My background is in software engineering and machine learning, and healthcare is one of the high-stakes domains where I have applied that work.

View Projects Download Resume ↓

San Diego · Los Angeles · New York City

github.com/KaustubhLall linkedin.com/in/kaustubh-lall kaustubhlall@gmail.com

project.map

cd ~/projects >_

DuskactiveAgent Systems · Healthcare MLResMed's public AI-powered sleep coach, with natural voice support, adaptive guidance, and personalized CBT-I plans. IP pending for the multi-agent CBT-I system.>_ cat dusk.md

Dino ArenaactiveGames & SimulationA reproducible, framework-independent simulation environment for fair agent benchmarking, not just a game.>_ cat dino-arena.md

Pulse CourtactiveGames & SimulationA from-scratch 1v1 arena game where fixed-point math replaces floats so the same input tape always reproduces the same state hash, built as a foundation for future RL, not a finished game.>_ cat pulse-court.md

NEAT PlaygroundactiveGames & SimulationNEAT neuroevolution implemented from scratch in C, evolving neural nets that play Flappy Bird, Snake, 2048, and Pac-Man, with interactive replay visualizers you can run in the browser.>_ cat neat-playground.md

AI BomberresearchGames & SimulationHigh-performance Bomberman-style AI simulation environment in C17, plus a native AlphaZero self-play campaign with honestly reported results, one retraction included.>_ cat ai-bomber.md

DawnactiveHealthcare MLResMed's AI-powered digital sleep-health assistant. I was one of the engineers who built it.>_ cat dawn.md

Drug Transporter MLarchivedHealthcare MLThe code behind a cited biomedical publication (Journal of Biological Chemistry, 2020).>_ cat drug-transporter-ml.md

OSRS Market Tools (MMG)activeML ResearchMoney-making-guide and Grand Exchange analysis for Old School RuneScape: deterministic margin analytics plus a leakage-checked ML research track on public market data.>_ cat mmg.md

Algorithmic SolvabilityresearchML ResearchControlled experiments on whether ML models detect that a task is governed by a compact algorithm.>_ cat algorithmic-solvability.mdLifelineactiveAgent SystemsA full-stack personal AI assistant with semantic memory, five chat modes, voice in and out, and Gmail tool calling via a LangGraph agent, deployed end to end on EC2.>_ view in listingAI FroggerresearchGames & SimulationFrogger AI simulation environment and visualizer, a companion benchmark to AI Bomber.>_ view in listingRingRoadactiveGames & SimulationA mobile extraction-shooter prototype for iOS, built in Unreal Engine 5.7 and C++.>_ view in listingCOMISA ModelingresearchHealthcare MLCurrent work: modeling for comorbid insomnia and sleep apnea (COMISA).>_ view in listingMSK ResearcharchivedHealthcare MLU-Net segmentation research for musculoskeletal MRI, a three-week summer exploration in Keras and TensorFlow.>_ view in listingPacMan AIarchivedML ResearchA from-scratch pygame Pac-Man testbed comparing NEAT, a prioritized-replay DQN, and an abandoned CoDeepNEAT/HyperNEAT detour.>_ view in listingRuneLite CV HelpermaintainedSystemsA 16.6k-line game-automation plugin decomposed into 22 bounded services, with a versioned HTTP contract and recovery ladders.>_ view in listingWhisperKeyarchivedToolsA Windows system-tray hotkey dictation tool with a cloud Whisper API default and an optional fully local engine, built but never merged out of a feature branch.>_ view in listingMeal TrackerarchivedToolsA Django and React meal-tracking app built in about nine days: REST API design, JWT auth, and database modeling.>_ view in listingHonestReviewerarchivedToolsA Django app that uploads Amazon review CSVs and flags fraud or low-helpfulness reviews via Cohere's classification API with hand-written few-shot examples.>_ view in listingChambal Awareness WebsitearchivedToolsA pro bono website for a river-restoration nonprofit in Kota, India, built in four days and deployed to Firebase.>_ view in listing

Projects

Case Studies

Publications

8.5+

Years of Experience

Featured Work

View all projects

Healthcare ML

Dawn

ResMed's AI-powered digital sleep-health assistant. I was one of the engineers who built it.

LLMs
Python
Cloud infrastructure
Conversational AI
LLM systems

Animated walkthrough of the live analyzer: the ranked money-making methods table with live Grand Exchange prices, then the GE Flipping margin scanner

ML Research

OSRS Market Tools (MMG)

Money-making-guide and Grand Exchange analysis for Old School RuneScape: deterministic margin analytics plus a leakage-checked ML research track on public market data.

Python
Pandas
OSRS Wiki API
Time-series ML
Evaluation methodology

Flappy Bird replay dashboard with live network visualization

Games & Simulation

NEAT Playground

NEAT neuroevolution implemented from scratch in C, evolving neural nets that play Flappy Bird, Snake, 2048, and Pac-Man, with interactive replay visualizers you can run in the browser.

C99
NEAT
SDL2
HTML/JS visualizers
Neuroevolution

whoami --verbose

I'm a software and machine learning engineer. I do not have a biomedical background; medicine has been an application domain for systems I know how to build. At UC San Diego, I applied software, classical ML, and computer vision to drug-transporter data and MRI segmentation, working with domain experts who supplied the biological and clinical context.

After a master's focused on machine learning, I spent about four years as a founding ML and cloud engineer at Radicle Science. I built data infrastructure, reporting systems, and recommendation software for a health-tech platform running direct-to-consumer studies. The engineering problems were broadly applicable: normalize messy data, make pipelines reliable, and turn model output into something people can use.

Today I build conversational AI and production ML systems at ResMed, including Dawn and Dusk. I also continued advising my former UCSD PI on medical-imaging work first written during my master's. Across healthcare, agents, research tooling, and simulation environments, the thread is software engineering: turn an ambiguous problem into a testable system, then find out what breaks when it meets real users, data, latency, and operational constraints.

Education

M.S., Machine Learning & AI
UC San Diego
2020 – 2021 · 1 year
Graduate study focused on software engineering and machine learning, with medical imaging as one application area.
The underlying research and manuscripts were completed in 2021. I continued advising my PI as the work progressed to publication in 2024 and 2025.
Research publications
- Segment Anything Model (SAM) and Medical SAM (MedSAM) for Lumbar Spine MRISensors · published 2025 ↗
- Shoulder Bone Segmentation with DeepLab and U-NetOsteology · published 2024 ↗
B.S., Computer Engineering
UC San Diego
2016 – 2020 · 4 years
Computer engineering foundation with software and ML research from second year, including applications to biomedical datasets.
Research publications
- Unique metabolite preferences of the drug transporters OAT1 and OAT3 analyzed by machine learningJournal of Biological Chemistry · published 2020 ↗

View all publications >_

Currently

Senior Machine Learning Engineer

ResMed · Feb 2024 – Present

Machine learning and conversational AI for sleep health, from GenAI platform infrastructure to consumer products and clinical modeling.

View experience >_

Kaustubh Lall _

>_Featured Work