Shyam KannanAI Engineer & Forward Deployed Engineer

I build AI tools around how teams work. My client projects span enrollment workflows and healthcare data, alongside independent work in agents and machine learning.

View client projects Get in touch

Client projects

Selected projects I worked on with clients.

Client project

EnrollIQ

An AI assisted workflow inside Zoho CRM that helps enrollment advisors review learner information and recommendations in the workspace they already use.

Read the EnrollIQ case study

Problem: Enrollment advisors handle learner inquiries and have to judge how interested a learner is, which courses fit, and which payment options may suit them.
My contribution: I spoke directly with the sales and enrollment team to understand how advisors worked before designing anything, then contributed to the roadmap, the Python backend, the React interface, and the OpenAI retrieval workflow.
Delivered: An AI assisted workflow within Zoho CRM that connects learner information to recommendations advisors review in their existing workspace.
Technologies: Python · React · OpenAI · Retrieval augmented generation · Zoho CRM

Client project

IVX Health

Healthcare data preparation, validation, and model evaluation for a team working on patient risk and visit prediction.

Read the IVX Health case study

Problem: A team working on patient risk and visit prediction needed healthcare records prepared and checked, and results they could inspect.
My contribution: I prepared and validated healthcare records and supported the team with model evaluation and dashboards that made results easier to inspect.
Delivered: Validated data for the prediction workflows, plus evaluation and dashboards that helped the team review model results.

Selected projects

RepoFriend

Point it at a GitHub repository and it clones the code, maps the structure, asks Claude to write documentation, and can open a pull request that adds inline comments.

The note walks through the LangGraph workflow step by step, how failures are recorded, what the pull request path does, and where the analysis stops.

Read the RepoFriend project note RepoFriend code on GitHub

LessGo

SJSU capstone project, built by a team

A carpooling app for SJSU students. A SwiftUI app for riders and drivers runs on Node.js and Python services, with student ID checks and ride matching that combines location filtering, learned embeddings, and a shared cost score.

The note separates my commits from the team's work and explains the ID check, the three matching stages, and their limits.

Read the LessGo project note LessGo code on GitHub

BioMed LLM

Answers yes, no, or maybe biomedical questions with BioGPT, adding context retrieved from PubMed through NCBI Entrez.

The note shows the retrieval and prompt flow from the notebook, how answers were scored, and why its accuracy output is not presented as a result.

Read the BioMed LLM project note BioMed LLM code on GitHub

About

I like work that starts with understanding a team's problem and ends with something they can use. On EnrollIQ, that meant speaking with the enrollment team and building around the CRM they already worked in.

My background is in computer science and artificial intelligence. Most of my recent work combines retrieval, LLM workflows, and the Python services around them, with a React or SwiftUI interface when a project needs one. Each project page on this site says what I did, how the system works, and where the evidence stops.

Education

M.S. in Artificial Intelligence, San Jose State University
B.S. in Computer Science, Arizona State University

Skills

AI and machine learning: Retrieval augmented generation · LLM workflows with LangGraph · Claude and OpenAI APIs · Hugging Face Transformers · Model evaluation
Backend and data: Python and FastAPI · Celery and Redis · PostgreSQL and PostGIS · Node.js with TypeScript
Interfaces and delivery: React and Next.js · SwiftUI · Docker · CRM integration with Zoho

Get in touch

Interested in my work? Email me or connect on LinkedIn.

shyamkannan.dev@gmail.com