Profile banner
Srihari Unnikrishnan

Srihari Unnikrishnan

Deep learning researcher and ML/AI engineer

srihari-unnikrishnan
Bengaluru, Karnataka, India
Joined February 2026
Favicon imagegithub.com
Favicon imagepythongiant.github.io
Favicon imageionio.ai
Favicon imagelinkedin.com

Network

2.7K connections

Summary

Applied machine learning researcher with a focus on financial and synthetic-data problems — contributed to a multi-author research paper applying Causal GANs and reinforcement learning to bond-yield forecasting and evaluation with LLMs. arxiv+1
Technical writer and educator on practical LLM tooling and inference — authored Ionio blog posts about LLM benchmarking and running quantized models locally, showing an emphasis on inference speed and accessible tooling. ionio+2
Active open-source contributor and developer — maintains GitHub projects and educational notebooks that explain transformer architectures and model-building from first principles. github+1
Broad industry experience spanning early-career internships to roles across consulting, research labs, startups, and Big Tech, reflecting a mix of applied engineering and research positions in ML and data science. weekday+2

Work

Education

Projects

Writing

Predicting Liquidity-Aware Bond Yields using Causal GANs and Deep Reinforcement Learning with LLM Evaluation

February 1, 2025

Research paper presenting a framework that uses Causal GANs and Soft Actor-Critic RL to generate synthetic bond yield data and a fine-tuned LLM for trading signals and risk assessments.

Favicon imagearxiv.org

Fastest Token First: Benchmarking OpenLLMs by inference speed

March 1, 2024

Blog post discussing latency metrics for LLMs, including Time To First Token (TTFT), and benchmarking inference speed of open models.

Favicon imageionio.ai

A Complete Beginners guide to localllm

January 1, 2024

A practical guide describing localllm tools to run quantized LLMs locally (including setup and hosting on Google Cloud).

Favicon imageionio.ai

Hobbies

Writing technical blog posts and contributing to open-source ML projects. github+2