also known as Rithesh R N
Senior Applied Scientist @ Microsoft
I build AI systems that ship — from research ideas to production products. My work lives at the intersection of LLMs, autonomous agents, and enterprise AI. I care deeply about research that matters and products that work.
San Francisco, CA
Always happy to grab coffee and talk AI, startups, or terrible puns.
I'm a Senior Applied Scientist at Microsoft in the Bay Area, working on the Sales Qualification Agent. My work spans model training, agent architectures, benchmarking, evaluation, and everything in between.
My long-term goal is to build systems that make people and organizations fundamentally better at what they do — technologies that sharpen judgment, unlock creativity, and raise the ceiling on what teams and individuals can achieve.
Previously, I was at Salesforce AI Research where I built AI systems across LLMs, agents, and enterprise products. Before that, I built conversational AI at Embodied Inc., did research with Prof. Julian McAuley at UC San Diego, and interned at Q-Centrix building ML models for healthcare. I hold an MS in CS (AI) from UCSD and a BE from R.V. College of Engineering, Bangalore.
Outside of work, I enjoy travelling and extreme sports.
20+ papers · 10+ patents · ICLR, NeurIPS, IEEE and more
An open-source framework for systematic, automated prompt optimization — making prompt engineering reproducible and scalable. Trusted by 900+ developers on GitHub.
View on GitHub →First comprehensive framework for evaluating LLMs and LMMs on mobile devices — spanning NLP, safety, and multimodal tasks across performance and efficiency dimensions.
Read paper →The model family that ranked #1 on the Berkeley Function-Calling Leaderboard, outperforming GPT-4 and Claude-3 on tool-use and agent benchmarks.
Read paper →Automated pipeline for verifiable, diverse function-calling datasets. Associated dataset xlam-function-calling-60k was Top-3 trending on HuggingFace in July 2024.
View on Scholar →Policy gradient-based framework that reinforces language agents through retrospective reflection. Selected as an ICLR 2024 Spotlight — top ~5% of submissions.
Read paper →Multi-agent framework that leverages the complementary strengths of diverse specialized agents to outperform any single framework on complex software engineering tasks.
Read paper →Full list on Google Scholar and OpenReview.
Microsoft · San Francisco, CA
Building the Sales Qualification Agent — applying LLM and agent expertise to transform enterprise sales workflows.
Salesforce AI Research · Palo Alto, CA
Built and shipped AI systems across LLMs, agents, and enterprise products. Work spanned research, engineering, and product integration — from concept to production at scale.
Embodied, Inc.
Built NLP models for text classification, question answering, and abstractive text summarization. Developed and deployed ML solutions for a social companion robot platform.
University of California, San Diego
Research on Personalized Dialogue Agents with Prof. Julian McAuley. Taught Data Science for Finance, Probability & Statistics, and Big Data Analytics using PySpark.
Q-Centrix
Built ML models predicting Cath PCI registry answers from Electronic Medical Records. Built deep learning extractive summarization models for EMRs using ELMo & BERT.
University of California, San Diego
R.V. College of Engineering, Bangalore
Travel, adventure, and a camera always within reach.
9 countries · 68 cities · always planning the next one.
I'm open to research collaborations, co-founding conversations, and advisory opportunities. If you're building something ambitious in AI — or just want to grab coffee in SF — reach out.