Vidharshana Ravi — hybrid search & GenAI systems for 1.5M+ users at Morgan Stanley. Published researcher. National hackathon winner. Founder on the side.
I'm a Search Engineer at Morgan Stanley, working across the full retrieval stack — ingestion pipelines, schema design, hybrid ranking, and the GenAI layer that turns search results into answers. The platform I help build serves 1.5M+ users globally, so most of my day is some mix of relevance tuning, reliability work, and figuring out why a query that should return one document returned zero.
Outside that, I build things on my own clock. I've shipped a full e-commerce platform from scratch, won national hackathons building accessibility and civic-tech tools, published peer-reviewed research on prosthetic-limb computer vision, and spoken to rooms of 100+ developers about generative models. I graduated from College of Engineering Guindy with a 9.0 CGPA.
Build hybrid retrieval — Solr/Fusion keyword search merged with vector database retrieval — and the GenAI answer-generation pipeline sitting on top of it, for an enterprise platform used by 1.5M+ people. Designed observability for response, summary, and interaction signals so the team can actually measure answer quality instead of guessing. Led an Elasticsearch modernization PoC and automated Solr backup validation, cutting manual verification effort. Promoted from Analyst (SDE1).
Researched grasp-pattern detection for prosthetic limbs using deep learning and computer vision (TensorFlow, Python) — accessibility-focused work that was later published in a peer-reviewed Springer CCIS volume.
CGPA 9.0. Deputy Director & Head of Content/Design at ACM, Joint Secretary at CSEA, Junior Head at AUSEC — running workshops, stock-market and crypto sessions, and technical content for student communities. Multiple national hackathon wins during this period.
An assistive communication system for people with disabilities — eye tracking, context recognition, and multi-modal input working together so users can communicate without relying on speech or fine motor control. Built and iterated across a 6-member team through multiple evaluation rounds before taking 1st place nationally.
view source ↗A multi-agent system for real-time city intelligence — civic issue monitoring, traffic analysis, geo-tagged incident reporting, and predictive alerts, built on Gemini, Vertex AI, and Firebase.
view source ↗A full-stack e-commerce platform I built and scaled from zero — product, deployment, customer acquisition, and go-to-market, end to end. Now serving 100+ customers and 700+ orders a month.
view source ↗An emergency-response and evacuation-optimization system built for war-time and disaster scenarios — routing people away from danger zones as conditions change in real time.
view source ↗An AWS-based self-service platform for Linux Kerberos, NFS, and Active Directory provisioning, plus storage and quota monitoring — replacing a manual identity/access workflow with automation.
Devfolio Women Hackathon (Top 3 of 500+ teams), IdeaCrunch, Innovision, HackIT/Innovathon (all 1st place), and CodeMan (3rd place).
see full list ↓Open to interesting roles, speaking invites, and collaborations. Fastest way to reach me is email.