Agentic AI & Orchestration
Multi-agent systems with LangGraph and MCP — planning, tool use, and task routing across models.

Lead AI Engineer, Kissflow — Chennai
I build production AI systems — multi-agent orchestration, RAG pipelines, and fine-tuned models — for teams that need them to hold up past the demo. Nine years across data engineering, backend systems, and now applied AI, most recently leading agentic and RAG architecture at Kissflow.
Day to day: LangGraph and MCP for agent orchestration, FAISS, Pinecone, and pgvector for retrieval, LoRA and QLoRA for fine-tuning open models, and the FastAPI, Python, and cloud plumbing that gets it all into production behind evals and guardrails. On the business side, I build GTM systems — document classification, quote-to-cash automation, and the ERP-adjacent workflow logic that makes AI actually useful inside revenue operations.
Shipped AI systems — with the metrics that made them worth building.

github.com/MayaDispeler/TheOrqestra
97 specialist agents. 101 skills. One orchestrator that assembles the right team before anyone writes a line of code.
Agent Orchestration · npm Package
github.com/MayaDispeler/b2b-intent-signal-classifier
Fine-tuning DistilBERT to detect 6 buyer intent signals from B2B texts — from synthetic data generation to Hugging Face deployment.
Fine-tuned LLM · Hugging Face
github.com/MayaDispeler/account-research-intel
Autonomous AI intelligence engine that turns a company domain into a 17-slide GTM briefing in 15 seconds — 14 data layers, zero hallucination.
Autonomous AI Agent
github.com/MayaDispeler/leadscore-ai
GPU-trained LightGBM model that scores B2B leads by conversion probability with SHAP explainability — AUC 0.83, trained on 300K synthetic leads.
ML · GPU Training
github.com/MayaDispeler/AI-Data-Analyst
Natural language to SQL interface — ask plain English questions, get back charts, tables, or JSON from MySQL, PostgreSQL, SQL Server, or Oracle.
NL-to-SQL · Multi-DBFrom agent orchestration and retrieval to fine-tuning and the production plumbing that keeps it all honest.
Multi-agent systems with LangGraph and MCP — planning, tool use, and task routing across models.
Vector search, hybrid retrieval, and reranking — FAISS, Pinecone, pgvector, knowledge bases.
LoRA, QLoRA, DPO, and PEFT on open models, benchmarked with BLEU, F1, and LLM-as-a-judge.
Context engineering, memory design, and token/cost optimization for long-running agents.
LangChain, LangGraph, DSPy, and MCP across OpenAI, Claude, and Gemini APIs.
Serving, observability, and guardrails — shipping agents and RAG pipelines that hold up in prod.

Kissflow Inc.
Leading AI initiatives and building intelligent products and go to market engineering at scale.
Neurealm (formerly GAVS | GS Lab)
Building data platforms and analytics solutions for enterprise clients.
TCS
Worked on core systems and enterprise applications.
Formal training across aerospace engineering and business administration.
2013 — 2017
SASTRA University

2024 — 2026
IGNOU University

A book, a blog, and a few things I've published along the way.

A Field Guide for Those Who Refuse to Shrink
Named after the honey badger — the animal that doesn't flinch — this book is for people who are done playing small. A sharp, honest field guide on thinking clearly, acting boldly, and building a life that doesn't bore you.
The people I've built with — from college to Kissflow and beyond.

“I'm writing to wholeheartedly recommend Srihari for any technology-related opportunity. I've had the pleasure of knowing him for nine years, since our college days, and have consistently been impressed by his leadership, technical expertise, and ability to leverage technology for success. From the very beginning, Srihari stood out as a natural leader. He possessed an uncanny ability to rally people around a common goal and bring out the best in everyone. Whether it was organizing college events or tackling complex projects, Srihari was always at the forefront, leading the charge with infectious enthusiasm and strategic thinking. But Srihari wasn't just a charismatic leader; he was also a tech whiz. He understood the power of technology to streamline processes, solve problems, and drive innovation. He readily adopted new technologies and wasn't afraid to experiment to find the best solutions. This tech-savvy approach was evident in all our college projects, where Srihari would seamlessly integrate technology to enhance our work and achieve outstanding results.”

“Srihari has always delivered solutions for the business problems and he takes care of the technology. He is a great asset for any business head looking to get insights for the business objectives.”

“Had the pleasure of working with Srihari at Kissflow. He was one of the best tech guys I have come across in my experience. We had worked together in a couple of projects across multiple domains. His enthusiasm for AI and early adoption has made him one of my go-to guys for tech doubts and queries. Amongst all his technical prowess he's a super supportive and focused individual. He always looks to rally the team to get the task done. Been a great pleasure working with him.”

“Srihari is someone you can always rely on when things get complex. A team with Srihari will benefit a lot at tough times and a team led by him will grow to their fullest potential. He has a strong grasp of AI and Technologies, but what stands out is how he applies it to solve real world problems that actually matter. He's always exploring and building, but with clear intent and not just for personal experimentation. He's also the kind of teammate who genuinely helps others, taking the time to share what he knows without hesitation.”
Have an idea, a problem worth solving, or simply want to connect? My inbox is open.
srihariv4942@gmail.com