AI Engineer · GenAI Platforms · AWS

Uwais
Mansuri

Building secure, production-grade GenAI systems and serverless platforms for regulated healthcare.

View Projects → Talk to Echo → View Resume ↗
alarmbrain — zsh
About

Who I am.

Uwais Mansuri
4+
Yrs Production
~3K
Daily Clinical Users
49
LLM Red-Team Tests
3
AI Tools Shipped

AI Engineer building a production GenAI service for Cigna Healthcare through Tata Consultancy Services, used daily by about 3,000 pharmacists and nurses. Before GenAI, I spent three years building and running the event-driven AWS platforms behind Cigna's healthcare applications, where reliability, observability, and secure PHI handling have real patient impact.

I pair deep cloud operations experience with a Master's in Data Science (GPA 3.9) to bridge the gap between raw infrastructure and intelligent systems. That combination drives the AI tooling I build on top of it — purpose-built tools that solve real problems I've lived in production.

I'm looking for AI Engineer, GenAI Platform, and AI infrastructure roles where secure, production-grade LLM systems matter. If you're building GenAI for healthcare or another regulated industry, let's talk.

🎓
M.S. Data Science
Saint Peter's University · Jersey City, NJ
2021 · GPA 3.9
🎓
B.Tech Computer Engineering
Indus Institute of Technology · Gujarat, India
2019
View Projects → View Resume ↗
Featured Work

Projects that ship.

cloudarchitect-ai — workspace $ generate: "e-commerce checkout · 10k RPS · HIPAA" Req Analyze Needs Arch Design System Diag Render Diagram Cost Estimate Spend Sec Audit Risks IaC Write Terraform GENERATED OUTPUT Services: ECS · ALB · RDS Aurora · ElastiCache · KMS Cost: ~$487/mo Security: 0 critical findings Terraform: main.tf · variables.tf · outputs.tf Powered by LangGraph · Amazon Bedrock (Claude)
Multi-Agent AI · LangGraph · Bedrock
CloudArchitect AI
Agentic AWS Architecture Generator

6-agent LangGraph pipeline that converts a plain-English prompt into a complete AWS architecture — diagram, cost estimate, security review, and production-ready Terraform in one shot.

LangGraph Amazon Bedrock FastAPI Lambda Terraform
Live Demo ↗ GitHub ↗
alarmbrain — incident brief
$ alarmbrain analyze --alarm quote-processor-throttles ✓ Classified: lambda_throttle ✓ Fetching domain context... ✓ Calling Claude API... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ INCIDENT BRIEF   INC-8FBF611F ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ Severity P2 Cause     Lambda concurrency limit hit Action    Check SQS queue depth ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
⚡ GenAI · Cloud · Healthcare
AlarmBrain
AI Incident Intelligence

Transforms raw CloudWatch alarms into structured incident briefs — probable cause, immediate actions, and healthcare business impact — in under 3 seconds across 8 AWS service types.

FastAPI Lambda Claude API SNS S3 Terraform
Live Demo ↗ GitHub ↗
clinicalscribe — encounter.mp3
▶ transcribing
→ SOAP
Patient Encounter — SOAP Note AI Generated
S Subjective: Shortness of breath on exertion, onset 3 days ago.
O Objective: BP 142/88, HR 94, SpO₂ 96%, lungs clear.
A Assessment: Hypertensive urgency, r/o cardiac etiology.
P Plan: Amlodipine 5mg, ECG, follow-up 48h.
🏥 GenAI · Healthcare AI · Bedrock
ClinicalScribe
AI Clinical Documentation

Converts patient encounter audio into structured SOAP notes using Amazon Bedrock and AWS MCP servers. Built with clinical UX standards and cost-safe feature flags for healthcare compliance.

Amazon Bedrock AWS Transcribe MCP FastAPI S3 Terraform
GitHub ↗
uwaismansuri.com Uwais Mansuri E Hi! I'm Echo. Ask me anything about Uwais...
Cloud + AI
Portfolio Website
This Site

Hosted on AWS S3 + CloudFront with Route 53 DNS. Features Echo, an AI assistant powered by Amazon Bedrock Claude via Lambda and API Gateway.

S3 CloudFront Lambda Bedrock
Live Demo ↗ GitHub ↗
etl-pipeline — running S3 Raw Data Lambda Transform RDS Store Orchestrated by Step Functions
Cloud Architecture
Serverless ETL Pipeline
Data Engineering

Automated data ingestion, transformation, and storage using Lambda, S3, RDS, and Step Functions with Parquet archiving for analytics.

Lambda S3 RDS Step Functions
Medium ↗
cnn-classifier — analyzing Medical Image CNN Layers Classified Melanocytic Nevus 84% confidence
Healthcare AI
Skin Disease Detection CNN
Computer Vision

CNN-based medical image classifier for dermatological conditions. Built with TensorFlow and Keras with a full image preprocessing pipeline.

TensorFlow Keras Python
GitHub ↗
← scroll →
AI Assistant

Ask Echo anything about Uwais

Powered by Amazon Bedrock · Claude. Ask about projects, experience, or open roles.

E

Echo

Uwais's AI portfolio assistant

◆ Amazon Bedrock · Claude
E
Hi! I'm Echo, Uwais's AI assistant. Ask me about his projects, experience, skills, or anything else you'd like to know.
Experience

Where I've worked.

Apr 2026 – Present
GenAI Platform Engineer
TCS  · Cigna Healthcare (Evernorth)
Current
LangGraph PydanticAI LiteLLM Anthropic Claude Bedrock AgentCore Lambda API Gateway DynamoDB KMS Terraform OpenTelemetry LangSmith
  • Architected and productionized an AWS serverless GenAI service that gives ~3,000 daily clinical users role-specific pre-encounter summaries from patient data via parallel retrieval, a de-identified LLM boundary, and deterministic post-processing.
  • Reduced generated output by about 700 tokens per request in a measured workflow through deterministic Python extraction; enabled prompt caching for reusable instructions and context to reduce repeated input processing and inference cost.
  • Designed a 49-case OWASP LLM Top 10-aligned adversarial and API negative-testing suite covering prompt injection, data leakage, unsafe output, resource exhaustion, schema validation, and JWT authentication; fixed a malformed-token crash and passed all defined scenarios using synthetic non-PHI data and live model calls.
  • Engineered deadline-aware execution with four parallel data fetches, concurrent LLM calls, timeouts, circuit breakers, and tiered fallbacks; corrected parallel tool-call output loss through exhaustive PydanticAI execution and deterministic merge and deduplication.
  • Codified GenAI infrastructure in Terraform across Lambda, API Gateway, IAM, KMS, Secrets Manager, networking, DynamoDB, CloudWatch alarms, and scheduled Provisioned Concurrency; delivered reusable CI/CD with structured logging and OpenTelemetry/LangSmith tracing.
Jan 2023 – Mar 2026
Cloud and Backend Platform Engineer
TCS  · Cigna Healthcare
Python Lambda API Gateway S3 SQS / DLQs DynamoDB Aurora PostgreSQL OpenSearch Comprehend EventBridge EKS CloudWatch Grafana Splunk Dynatrace
  • Owned development and production support for an event-driven AWS platform shared by four enterprise applications, using Python, Lambda, API Gateway, S3, SQS/DLQs, DynamoDB, Aurora PostgreSQL, OpenSearch, and Amazon Comprehend.
  • Automated email intake from Pega by processing .eml files and attachments through S3, SQS, and Lambda; invoked a custom Comprehend model to classify RFP and non-RFP submissions, deduplicate events, and route downstream workflows.
  • Operated CDC and transformation pipelines that normalized JSON events and replicated data from raw to integrated tables; diagnosed Lambda failures and database check-constraint violations that interrupted replication.
  • Sustained 99.9% platform availability and met 98% of operational SLAs by tuning Lambda concurrency and throttling, resolving SQS delays and DLQs, and monitoring CloudWatch, Grafana, Splunk, and Dynatrace synthetics.
  • Led EKS release validation and pod recovery during production deployments; coordinated incident RCA and implemented VPC, subnet, and load-balancer improvements to strengthen release stability and recovery.
  • Secured API access with a scheduled Okta token-refresh workflow using EventBridge, Lambda, Secrets Manager, and DynamoDB, plus group-based authorization through a Lambda Authorizer.
Feb 2022 – Dec 2022
Data Engineer
TCS  · Internal data platform
Python SQL Dimensional Modeling ETL Data Quality
  • Designed dimensional data models (fact and dimension tables) and built Python and SQL ETL jobs with data-quality checks for an organization-wide data warehouse feeding business reporting; defined data-mart standards and reviewed designs, code, and test plans.
Sep 2020 – May 2021
Applied Data Science Project
Saint Peter's University  · Faculty-led Kmart project
Python Pandas scikit-learn NLP
  • Built a customer-review sentiment classifier using Python, Pandas, and scikit-learn, identifying positive and negative feedback and operational themes such as delivery speed and logistics quality.
Capabilities

Tech Stack

☁ Cloud & Infrastructure
Lambda EC2 RDS S3 API Gateway SQS DynamoDB SNS CloudWatch IAM VPC Route 53 CloudFront ECS Kinesis Secrets Manager
🤖 AI & GenAI
Amazon Bedrock Claude API Prompt Engineering MCP Servers AWS Transcribe LLM Integration
👁 Observability
Grafana Dynatrace APM Dynatrace Synthetics Splunk CloudWatch ServiceNow X-Ray
⚙ IaC & DevOps
Terraform CloudFormation Docker Git CI/CD
</> Languages & Frameworks
Python SQL Bash FastAPI Pydantic Boto3
Credentials

Certifications

Let's Connect

Get in Touch

Let's work together.

Open to AI Engineer, GenAI Platform, and AI infrastructure roles where secure, production-grade GenAI systems matter.

LinkedIn ↗ GitHub ↗ uwais22@gmail.com