Sharmila Begum
Agentic AI Developer at Edgematics Group · Hyderabad, Telangana, India
As an Agentic AI Developer, I am driven by a profound curiosity to master complex systems and build innovative solutions that deliver tangible results. My pragmatic approach, coupled with a strong sense of accountability, ensures I consistently create robust and efficient AI architectures, always striving for mastery in my craft.
Experience
Agentic AI Developer
Devpost · Jul 2026 – Sep 2026
- I designed and built CareCircle end to end — from the core AI agent and clinical logic to the interfaces, deployment, and documentation.
- Below is a breakdown of what I personally contributed.
- AI Agent & Orchestration
- Architected the conversational agent using the Strands Agents framework with AWS Bedrock (Claude) as the reasoning engine.
- Authored the agent's system prompt and behavior guidelines — empathetic tone, evidence-based responses, and strict guardrails that keep it a screening-coordination tool rather than a diagnostic one.
- Designed the tool-orchestration layer so the agent can autonomously select and chain the right tools (risk scoring, scheduling, care planning, education, notifications) based on natural-language input.
- Clinical Logic & Custom Tools
- Implemented a risk assessment engine based on a modified Gail Model that scores breast cancer risk across 12+ genetic, lifestyle, medical-history, and reproductive factors, then classifies patients into risk categories with recommendations.
- Built the screening scheduler (mammograms, MRIs, ultrasounds, clinical exams) with preparation instructions and appointment tracking.
- Developed the care plan generator that produces personalized, risk-tiered plans covering screening schedules, lifestyle recommendations, action tasks, and genetic-counseling triggers.
- Created the patient education library (10+ evidence-based topics) and a notification/reminder system to improve screening adherence.
- Application & Interfaces
- Built three ways to interact with the agent: an interactive Streamlit dashboard with Plotly visualizations, a FastAPI REST API with full OpenAPI docs, and a CLI chat interface.
- Designed a graceful demo mode that runs the full workflow (risk → schedule → care plan → education → reminders) without requiring AWS credentials, so reviewers can try it instantly.
- Engineered the tools with an optional-dependency fallback so the UI runs even when the heavy AWS agent stack isn't installed.
Agentic AI Developer
Edgematics Group · Feb 2022 – Jun 2026
- Architecting end-to-end IDP pipeline using AWS AI services (Textract, Comprehend,
- Bedrock) to automate document processing at scale, reducing manual effort by 80%.
- Implementing Amazon Bedrock with Claude v2 and Titan LLMs for document
- classification, RAG-based Q&A, and entity extraction using few-shot prompting.
- Deploying Amazon Textract APIs to extract structured/unstructured data from invoices,
- forms, and identity documents with 95%+ accuracy.
- Building RAG pipelines using LangChain, AmazonTextractPDFLoader, and FAISS vector
- database for context-aware document retrieval and generative AI workflows.
- Training custom Amazon Comprehend classifiers and NER models with real-time
- endpoints to extract business-specific entities from multi-format documents.
- Developing Python automation scripts using boto3 SDK for document preprocessing,
- chunking, embedding generation, and AWS service orchestration.
- Provisioning serverless IDP infrastructure using AWS Lambda, S3, CloudFormation IaC
- templates, CloudWatch monitoring, and IAM least-privilege policies.
- Implementing PII/PHI redaction using Comprehend custom entities and integrated
- Amazon A2I for human-in-the-loop review workflows.
- Applying LLMs for document summarization, spellcheck correction, template-based entity
- extraction, and in-context Q&A with structured JSON output.
- Deploying FAISS vector store with Amazon Titan embeddings for semantic search,
- similarity matching, and MMR retrieval in RAG architectures.
- Contributing reusable production-grade Python code to AWS sample repositories
- demonstrating IDP patterns with Textract, Bedrock, and LangChain.
- Enforced AWS security best practices with IAM role-based access, data encryption, and
- HIPAA-compliant PII handling for healthcare document processing.
- Optimized LLM token usage and API costs through efficient prompt engineering,
- asynchronous Textract processing, and CloudWatch performance monitoring.
Gen AI Developer
Edgematics · Feb 2022 – Apr 2023
Software Developer
Mahendrada solution pvt ltd · Jul 2018 – Feb 2022
Skills 0 proven through work
Also works with
Proof of Work
Proof of Work
Sharmila shares this with people who ask. You'll hear back either way.
Education
DMSSVH College of Engineering, A.P, Artificial Intelligence
Krishna University, Machhlipattanam · 2012 — 2015
Contact details
Contact details
Sharmila shares this with people who ask. You'll hear back either way.
What drives their work
The Builder
You turn complex requirements into robust, functional systems.
Sharmila's superpower
You excel at constructing comprehensive AI-driven pipelines and applications from concept to deployment.
How Sharmila works
You are well-suited for roles on teams that require hands-on development of AI/ML systems and a focus on practical implementation.
Ask about this candidate
AI responses may contain errors. Proof status and source information are provided by Proof of Skill.
A Proof CV — one profile, kept current, with the proof attached. Back to the portal