Sreenidhi Vedartham AI Systems Builder • Embedded/IoT Engineer • AI Enablement

I build and ship AI systems that move from prototype to something a team actually uses. I turn LLMs into agents that are reliable and measurable, and I'm honest about where they save real time versus where it's just hype.

My background is embedded systems, IoT, and technical program management. The last few years I've shifted hard into AI: building tools, running experiments, and helping engineering teams tell the difference between real gains and noise.

AI Enablement Program, Ivani (IoT/OT) Designed role-specific prompt libraries, workflows, workshops, and adoption practices.

Results

• Organization-wide AI adoption in 2 weeks
• ~50% reduction in engineer rework
• ~8 minutes saved per AI query
• Reusable libraries now used across engineering, business and marketing teams

Stack: Hermes (Fireworks AI), GLM, MiniMax M3

AI Governance, Ivani Built practical guidelines, evaluation frameworks, and review processes for responsible AI use in engineering and marketing workflows.

TeamTrail, Slack Agent Builder Challenge Stateful Slack onboarding agent that lives in the top bar, not a slash command. New hires get personalized briefings grounded in real workspace messages (Slack RTS API) and Notion docs (via MCP), with every claim cited back to its source. Built smoke tests early to catch API/MCP schema mismatches before full integration. Stack: Bolt JS, Groq (LLaMA 3.3 70B), Slack RTS, MCP (Notion)

Memory-Augmented AI Agent, Global AI Hackathon Persistent memory layer that handles updates, conflicts, and forgetting instead of just appending facts forever. LLM-arbitrated conflict resolution with weighted dedup and importance-ranked recall. Deployed to cloud, shipped solo in one week. Stack: FastAPI, PostgreSQL/pgvector, Redis

Dead Code Finder, GitLab Transcend Hackathon GitLab Duo agent that answers "is anyone actually calling this," with honest confidence levels (Confident / Uncertain / Skipped) instead of overclaiming. Read-only by design, it reports, it doesn't modify anything. Validated two independent ways and confirmed two genuinely dead SDK methods by both routes. Stack: GitLab Duo Agent Platform, GitLab Orbit, Python

• CompassionateConnect - Multi-agent mental health intake platform using Gemini 1.5 and Firestore with crisis detection and therapist-ready summaries.
• GreenGrid - Energy forecasting platform built with IoT data, SageMaker, and Bedrock.
• Therapist Dashboard - Claude-powered clinician workflow assistant validated with feedback from a practicing therapist.

8+ years coordinating hardware, embedded, cloud, and data teams in IoT environments, sitting at the intersection of engineering execution and business outcomes. I help teams turn emerging AI capabilities into repeatable, measurable workflows.

Earlier in my career I worked in embedded systems at Trane while completing my M.S. in Electrical Engineering at the University of Texas at Tyler.

I co-founded and led Embark Women (2017-2022), growing an 800+ member community and personally mentoring 150+ women, with 20+ programs serving 500+ participants.

AI Enablement & Adoption, Agentic AI Systems, Technical Program Management, Workflow Automation, AI Governance, Multi-Agent Architectures, Model Context Protocol, AWS Bedrock, Gemini, Qwen, Hermes/Fireworks, LLM Evaluation & Prompt Engineering

H-1B transfer only. Approved I-140, no lottery required.

LinkedIn · Devpost · GitHub