Sachin Prabhakar

Software Engineer (ML/DL, DS/Infra) | 6+ YOE | 2x First Author ML Research Publications

San Francisco Bay Area, CA Santa Clara University — MS Computer Science & Engineering

I build machine learning systems and the distributed infrastructure that keeps them fast under load. At Santa Clara University I am a graduate researcher in the Machine Learning & Safety Analytics Lab, where my work fuses wearable sensing with generative AI to make occupational risk both measurable and explainable.

Before graduate school I spent six years shipping backend and data platforms in production: high-throughput payroll and computation engines on AWS Batch and sharded MongoDB serving 1M+ active users, Kafka event backbones with dead-letter recovery, and latency work that took hot paths from 12s to under 200ms. I care about the same thing in both worlds: systems that stay correct and legible at scale.

Google Scholar GitHub LinkedIn ORCID ResearchGate X

6+ yrsProduction engineering across ML, data, and infrastructure

2xFirst-author publications (HCII 2026, NAMRC 54)

400K+Wearable IMU samples modeled at 91% accuracy

1M+Active users served, load-tested to 10M

01 / Research

Publications

First-author work in wearable sensing, generative AI, and ergonomics analytics.

HCII 2026 Springer Nature First author 2026

EMG-Aware Generative AI for Explainable Ergonomics Risk Narratives in Repetitive Lifting

Sachin Prabhakar, Fatemeh Davoudi Kakhki

Digital Human Modeling and Applications in Health, Safety, Ergonomics and Risk Management · HCII 2026 Proceedings

A novel framework fusing wearable EMG sensing with generative AI to produce explainable ergonomic risk narratives for workplace safety, turning raw muscle-activation signals into reviewable, human-readable assessments.

Electromyography Generative AI Explainability Occupational Safety Google Scholar

NAMRC 54 Manufacturing Letters · Elsevier First author 2026

Wearable-IMU Machine Learning for Exertion and Fatigue Assessment with an Occupational Exoskeleton in Manufacturing

Sachin Prabhakar, Armin Moghadam, Fatemeh Davoudi Kakhki

54th North American Manufacturing Research Conference · Manufacturing Letters (Elsevier, 2026)

91% task-intensity classification accuracy across 400K+ IMU sensor samples, quantifying worker fatigue and enabling adaptive control of occupational exoskeletons on the factory floor.

Wearable IMU Time-Series ML Exoskeletons Human Factors ORCID record

02 / Experience

Work

Research, applied ML, and large-scale backend systems.

Research Assistant

Machine Learning & Safety Analytics (MLSA) Lab, Santa Clara University

Apr 2025 – Present

Built ML models on 400K+ wearable IMU sensor samples, reaching 91% task-intensity accuracy (NAMRC 54).
Engineered an EMG-aware generative AI system that produces explainable ergonomics narratives (HCII 2026).
Designed a real-time computer vision pipeline (MediaPipe) for worker posture detection and REBA-scored ergonomic evaluation.

Student Intern

Miller Center for Global Impact

Oct 2024 – Mar 2025

Built a mentor-entrepreneur matching engine using GPT-4 and SBERT embeddings over LinkedIn and Salesforce records, at 90%+ match accuracy.
Developed a GPT-4 pitch-deck screening pipeline with OCR and semantic segmentation, cutting review time by 70%.
Architected RAG pipelines (Sentence-BERT + FAISS) evaluating 1,000+ venture profiles with 93% classification accuracy.

Software Development Engineer III

BetterPlace

May 2021 – Aug 2024

Architected high-throughput payroll and computation engines on AWS Batch and sharded MongoDB, serving 1M+ active users and load-tested to 10M.
Built a Kafka pub-sub event backbone with dead-letter queues, holding sync errors under 0.5% across enterprise services.
Reduced latency from 12s to under 200ms (98% drop) on high-traffic services through Redis caching and query optimization.
Built real-time EWA microservices (Django, Celery, PL/SQL) accelerating loan go/no-go cycles by 60%.

Senior Software Developer

Navjoy Inc.

Nov 2019 – Apr 2021

Engineered a multi-tenant RBAC engine with JWT token lifecycles and fine-grained data-level policies.
Deployed automated open-data ETL pipelines using MongoDB and Cronhub, reducing manual workflows by 80%.
Integrated Amazon Rekognition CV models to automatically detect and index municipal field infrastructure.

Software Engineer

MetricStream

Aug 2016 – Jul 2017

Optimized enterprise Policy Document Management APIs, cutting read latencies by 50% with Redis.
Implemented PL/SQL data retention and cleanup routines, trimming table overhead by 40%.

03 / Stack

Technical skills

Core domains

Distributed Systems Large Language Models ML Infrastructure Generative AI Wearable Sensing Computer Vision

Languages & frameworks

Python PyTorch LangGraph Django Flask FastAPI TypeScript React.js

Data & cloud

Kafka Redis MongoDB PostgreSQL AWS Batch S3 Docker Prometheus New Relic

04 / Contact

Let's build something

Open to conversations on ML systems, distributed training, and research collaborations in wearable sensing and applied generative AI.

Email LinkedIn Scholar