Mehdi Salhi
AWS Cloud
Engineer
Computer Science student graduating August 2026 (SNHU). Building serverless apps and AI agents on AWS. Based in Stamford, CT / NYC.
Unkommon.ai
Built
AWS
Cloud Platform
3.71
GPA @ SNHU
About
Computer Science student at Southern New Hampshire University, graduating August 2026 (GPA 3.71). I build on AWS: serverless apps, RAG systems, and voice agents using Lambda, Bedrock, DynamoDB, API Gateway, Cognito, and AWS SAM. AWS Certified Cloud Practitioner (May 2026); pursuing Solutions Architect Associate (June 2026) and HashiCorp Terraform Associate (late June 2026).
I founded Unkommon.ai to solve a problem I kept seeing: companies spending months on AI projects that never make it to production. The model works in a notebook but breaks on real data. The chatbot demo impresses leadership but can't handle actual customer questions. Unkommon builds the systems that actually ship. On your infrastructure, with your data, in weeks.
On the side I'm wrapping up my Computer Science degree at Southern New Hampshire University, graduating August 2026 with a 3.71 GPA. School has been good but honestly building things has taught me more.
3.71
Dean's ListGPA
Southern New Hampshire University
Aug 2026
Graduating
Computer Science, B.S.
CT / NYC
Location
Stamford, Connecticut
AI + Cloud
Focus
AWS, LLMs, Agentic Systems
Skills
Technologies I work with
AWS & Cloud
AI / ML
Software
Certifications
Cloud and infrastructure credentials
In Progress
Solutions Architect
Terraform Associate
AI Practitioner
Earned
Cloud Practitioner
More credentials being added throughout 2026
Projects
Selected work and coursework
Unkommon.ai
Solo-built AI services platform on AWS. Voice receptionist (Vapi + Bedrock + Lambda) and website chatbot that qualify leads and book appointments via Google Calendar. Fully serverless. Lambda, Bedrock (Claude Sonnet 4.5 / Haiku 4.5), DynamoDB, API Gateway, Cognito, Amplify. React + TypeScript frontend, deployed via AWS SAM.
Company Policy RAG System
Production-grade RAG system with hybrid search, cross-encoder reranking, and RAGAS evaluation metrics. Upload any document and ask questions in natural language. Built with Python, LangChain, FastAPI, deployed on Hugging Face.
Bees Knees AI
Live marketing site for an AI agency with an embedded chatbot (Buzz) built on Claude Sonnet 4. Streaming SSE, prompt caching for ~90% cost reduction on repeated requests, per-IP rate limiting, and full security headers (HSTS, CSP, Permissions-Policy). Includes Cal.com booking integration for inline conversion and a custom WebGL shader hero.
MuscleMeal AI
iOS app that generates personalized meal recipes using Gemini AI, tailored to fitness goals and dietary preferences.
Get in Touch
Always open to new opportunities and conversations