Skip to content
View vineethsaivs's full-sized avatar

Block or report vineethsaivs

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
vineethsaivs/README.md
Vineeth Sai · AI/ML Engineer · Agentic Systems · Cosmos Dreamer


🌌 whoami

The universe is 13.8 billion years old. Somewhere along the way, atoms arranged themselves into things that could ask questions about other atoms. I think that's the wildest fact in existence, and most of what I do is downstream of that wonder.

By day, I build agentic AI workflows at AT&T: multi-process routing, computer-use agents, LangGraph pipelines that turn screen recordings into executable automations. On the side, I lead AI/ML at Nabu Tutor, an AI-powered USMLE prep platform, where I work on agentic RAG, long-term memory systems, LLM fine-tuning, and voice tutoring. Before all this: Google AgentSpace, Foundation AI, Aion Labs.

When I'm not wiring LLMs into production, I'm probably reading about black holes, watching City fall apart in a winnable match, or losing a chess game I should've won.


🛰️ Current orbit

location:    San Francisco, CA 🌁
role:        AI/ML Engineer @ AT&T  ·  Lead AI/ML @ Nabu Tutor
focus:       agentic systems · LLM post-training · long-term memory · RAG
research:    GRPO · RL for agents · multimodal reasoning · computer-use
open_to:     agentic AI collabs · research convos · hackathon teams

🧠 The arsenal

LLMs · Agents · RAG

LangChain LangGraph LangSmith OpenAI Anthropic MCP Mem0

Deep Learning · Fine-tuning

PyTorch HuggingFace Unsloth QLoRA PEFT TensorFlow

Vector · Memory · Data

pgvector Redis PostgreSQL Pandas NumPy

Cloud · MLOps

GCP Azure AWS Docker GitHub_Actions

Frameworks · Voice · Automation

FastAPI Playwright ElevenLabs Streamlit

Languages

Python TypeScript SQL Bash


📡 Signals from the observatory


🪐 Beyond the keyboard

Stuff that makes me a person, not a model.

  • 🌌 Cosmos: astrophysics, black holes, exoplanets, the early universe. If AI didn't exist, I'd be doing this.
  • 💙 Manchester City: through the trebles, the heartbreaks, the existential VAR decisions. Cityzen for life.
  • 🎾🏎️ Tennis & F1: weekend rituals. Slams and Sundays.
  • ♟️ Chess: vineethsaivs on chess.com. Down for a game anytime.
  • 🍥 Naruto: "Hard work is worthless for those that don't believe in themselves." Believer.
  • 🥷 SF hackathon scene: Zero to Agent, OpenEnv, Shack15 regular. Built a VLM/GRPO Duck Hunt agent that hit 60.9% on Horizon-Min. Always down to team up.

🌠 Connect


"Somewhere, something incredible is waiting to be known."
- Carl Sagan

⭐ Made with cosmic curiosity and probably too much coffee.

Pinned Loading

  1. DeepSpeed DeepSpeed Public

    Forked from deepspeedai/DeepSpeed

    DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.

    Python 1

  2. pytorch pytorch Public

    Forked from pytorch/pytorch

    Tensors and Dynamic neural networks in Python with strong GPU acceleration

    Python 1

  3. unsloth unsloth Public

    Forked from unslothai/unsloth

    Unsloth Studio is a web UI for training and running open models like Gemma 4, Qwen3.6, DeepSeek, gpt-oss locally.

    Python 1