Education
M.S. Computer Science (AI/ML)
Northeastern University
About
Harshil Chudasama
I build AI-powered products and the backend systems behind them.
Real-time computer vision, recommendation systems, search and ranking, and the low-latency infrastructure that serves them. I work across the full cycle: model optimization and on-device deployment, event-driven architectures at scale, and production performance tuning.
Before engineering, I worked in product management and business analysis, which is why I think in user outcomes and tradeoffs, not just system diagrams.
What drives me
Products where ML reaches users at scale.
Core stack
PythonJavaPyTorchAWSKafkaSpring BootElasticsearchDockerReactTypeScriptPostgreSQLRedisPythonJavaPyTorchAWSKafkaSpring BootElasticsearchDockerReactTypeScriptPostgreSQLRedis
Outside of work
Training at the gym
Building a jewelry brand
Reading about markets
Exploring new cities
Tennis
ML + AI
PyTorchONNX RuntimeTransformersYOLOv8Elasticsearch kNNRAG
Backend + Infra
JavaSpring BootFastAPIKafkaPostgreSQLAWSDocker
Product + Systems
Model ServingSearch RankingLow LatencySystem DesignAPI Design
Projects
Selected Work
Personalized Recommendation System
CineMatch
Measurable MRR and NDCG gains for personalized recommendations. Two-stage retrieve-then-rank engine with collaborative filtering, Elasticsearch kNN vector retrieval, and Redis-cached user embeddings.
Read more →
Hybrid Search Engine
Semantic Search Service
Improved search relevance with measurable MRR and NDCG gains. Hybrid ranking combining BM25 lexical retrieval with dense vector embeddings via Elasticsearch kNN.
Read more →
Deadline-Aware Model Serving
Anytime Inference
Maintained sub-50ms p99 latency for 500+ concurrent users. Deadline-aware inference router that switches between FP32 and INT8 based on CPU pressure.
Read more →
Browser-Based Virtual Try-On
FaceFit AR
Sustained 60 FPS face tracking in the browser. Real-time eyewear try-on using MediaPipe Face Mesh with 468 3D landmarks and a Three.js rendering pipeline.
Read more →
Event-Driven E-Commerce Backend
Commerce Services
Cut deadlocks by 40% under fault injection. Event-driven e-commerce backend built with Saga orchestration and optimistic concurrency control.
Read more →
Algorithmic Trading Engine
Execution Copilot
Improved simulated execution quality under volatile conditions. Low-latency trading engine with contextual bandit policy selection and Monte Carlo simulations.
Read more →
Mobile Identity Verification
QuickVerify SDK
Production-ready verification SDK for iOS and Android. Cross-platform React Native package with native biometrics and document capture modules.
Read more →Experience
Experience
SEP 2025 - PRESENT
Vancouver, BC
Software Developer Intern
ICBC
APR 2025
Hackathon
Microsoft x Qualcomm On-Device AI Hackathon
AwardMicrosoft x Qualcomm
1st Place
Real-time on-device navigation for wheelchair users. YOLOv8 and Whisper on Snapdragon X Elite, sub-40ms latency via DSP/NPU offloading.
MAR 2025 - JUN 2025
Vancouver, BC
AI Research Assistant
Northeastern University, Mixed Reality Lab
SEP 2024 - DEC 2024
Vancouver, BC
Graduate Teaching Assistant
Northeastern University
AUG 2024
Global
IMC Prosperity 2 Trading Competition
AwardIMC Trading
Rank 87 / 13,500 (Top 1% Global)
Autonomous trading bot with Monte Carlo simulations and dynamic inventory management.
MAY 2024 - AUG 2024
Boston, MA
Software Engineer Intern
Thermo Fisher Scientific
Contact
Get in touch
Currently
Interested in AI-powered products, personalization, and discovery.