Founding Engineer (Core Team)
Moss
San Francisco, CA
Oct 2025 – Present
First core-team hire: built the Rust hybrid search engine (BM25 + semantic), shipped at 4ms P99 and now behind 300k+ SDK downloads.
Visit Company PageAI/ML Engineer · San Francisco
I am Yatharth
Innovating AI for smarter solutions and scalable systems.
About Me
I'm a software engineer in San Francisco building production AI systems: hybrid search in Rust, RAG pipelines, and real-time voice agents. I care most about the gap between an ML demo and a system people rely on: retrieval quality, latency, and cost.
Before this: an MS in Computer Science from Indiana University Bloomington, LLM-security research at Kelley's DSAIL lab, and engineering stints across AI startups. The chatbot on this site, Inpersona, is my own RAG stack, self-hosted end to end on a $5/month server.
Experience
Software Developer
Haldune
Chicago, IL
Jun 2025 – Oct 2025
Cut agentic research-report generation from 10 minutes to under 3 with a parallel multi-agent workflow on Azure AI Foundry.
Visit Company Page
Research Assistant
DSAIL, Kelley School of Business
Bloomington, IN
Mar 2024 – Dec 2024
Red-teamed LLMs with Garak, Counterfit, and Purple Llama; NeMo guardrails cut attack success rate by 60%.
Visit Company Page
Software Engineer Intern
Outspeed
San Francisco, CA
Jun 2024 – Aug 2024
Built a GPU speech-to-speech stack (Whisper, Llama 3.1, Parler TTS) on AWS EKS at p95 800ms; self-hosted vLLM cut inference costs ~30%.
Visit Company Page
Software Developer Intern
IDeaS
Pune, India
Jul 2022 – Jan 2023
Built an analytics platform for five multi-billion-dollar hospitality clients, replacing Excel macros and saving 100+ engineering hours a month.
Visit Company Page
Software Engineer Intern
Quidich Innovation Labs
Mumbai, India
May 2022 – Jun 2022
Pose-estimation algorithm predicting bowler release points at 97% accuracy, powering live broadcast analytics on Sony Liv.
Visit Company Page
Software Developer Intern
Azodha
New York, NY
Sep 2021 – Apr 2022
Led Open-Notif: event-driven notifications on AWS with sub-3-second delivery and 99% production uptime.
Visit Company PageProjects
Image Inpainting
Reconstructs masked facial regions with partial convolutions in PyTorch, trained on 50,000 CelebA images with random-walk-generated masks to prevent overfitting.
Sprechen
Transformer-based machine translation implemented from scratch in PyTorch: English to Hindi and German to English.
Highway Agent
A reinforcement learning agent for autonomous highway navigation, trained with an actor-critic policy-gradient setup.
PixTOPix
A conditional GAN that translates satellite imagery into map tiles, trained on a 100K-image Google Maps dataset.
Open-Notif
Event-driven notification CLI on AWS (Lambda, SNS, SQS) with PostgreSQL, Twilio, and Mailchimp. Sub-3-second delivery, with infrastructure managed as code via AWS Chalice.
Skills & Expertise
Languages
Rust
Python
C++
Go
TypeScript
JavaScript
SQL
Technologies
PyTorch
TensorFlow
Django
Node.js
Express
React
Next.js
Tailwind CSS
FastAPI
Docker
Kubernetes (EKS)
AWS
Kafka
Git
GraphQL
Databricks
Spark
Airflow
GitHub Actions
Databases
MongoDB
Neo4j
DynamoDB
Redis
Core Competencies
Hybrid Search & Vector DBs
LLM Engineering
Real-Time Systems
ML Model Finetuning
AI Integration in Web Apps
AWS Cost Optimization
CI/CD
Achievements & Awards
Co-led & judged two 200-person hackathons at Y Combinator HQ
2025 & 2026
Co-led and judged two 200-person hackathons hosted at Y Combinator headquarters in San Francisco, evaluating startup demos and mentoring builders across both cohorts.
Graduate Teaching Assistant, Indiana University
Aug 2024
Selected as a Graduate Teaching Assistant for Elements of AI (with Prof. David Leake) and Machine Learning in Signal Processing (with Prof. Jonathan Ponniah), supporting graduate students with lectures, assignments, and project mentorship.
SIH 2022 Grand Finale Winner
Aug 2022
Led the creation of a no-code chatbot-building platform at the SIH hackathon, competing among thousands of teams. Enhanced chatbot narrative understanding with AWS ECS, implemented Amazon S3 for conversational storage.
4th Rank in India Academia Connect AI Hackathon
Oct 2021
Created AI solutions for COVID-19 distancing, using YOLO for crowd monitoring and heat maps for sanitation planning.
87th Rank in Amazon ML Challenge
Aug 2021
Ranked 87th of 3,300 teams in Amazon's ML Challenge with a TF-IDF product-categorization model.
Testimonials
“I hired Yatharth as the first engineer at Moss, and he quickly became the person I'd hand the hardest problem to and stop worrying about: relentlessly hardworking, completely dependable, with work that always came back done, and done well. He owns problems end to end and learns fast, but what really sets him apart is that he leaves everything he touches better than he found it and holds himself to a higher bar than anyone asks of him. He's sharp, low-ego, and a real pleasure to build with. He'll make any team better, and I recommend him without reservation.”
Get in Touch
Have a project in mind or want to discuss potential opportunities? I'd love to hear from you!