Skip to content

Divya Burugu

AI/ML Engineer • Atlanta, GA • drivetube.ai/•••••

Professional Summary

AI/ML Engineer with 5+ years of experience architecting enterprise-grade machine learning systems, LLM-powered applications, and scalable data pipelines across healthcare, financial services, and telecommunications. Deep expertise in Retrieval-Augmented Generation, semantic search, and LLM orchestration using LangChain, FAISS, HuggingFace Transformers, and OpenAI GPT-4. Proven record designing production-ready FastAPI microservices with OAuth 2.0 security, Pydantic validation, and fallback mechanisms that satisfy enterprise reliability and compliance standards. Skilled in end-to-end MLOps from XGBoost and Scikit-learn model training and SMOTE-based class balancing to MLflow experiment tracking, batch scoring automation, and CI/CD deployment via GitHub Actions and Azure DevOps. Experienced monitoring production models for feature drift, prediction distribution shifts, and delayed-label evaluation in regulated environments, applying SHAP and LIME for audit-ready explainability. Strong hands-on background with AWS SageMaker, EC2, S3, Lambda, and Azure ML for cloud-native model deployment and retraining pipelines.

Technical Skills

Programming Language: Python,K
Backend Technologies: FastAPI,Flask,Pydantic,Retry Logic,Timeout Handling,REST APIs,OAuth 2.0,PyTest
Cloud Platforms: AWS
DevOps & Infrastructure: Azure DevOps,Docker,GitHub Actions,CI/CD Pipelines
Data Analysis & Visualization: Pandas,NumPy,PostgreSQL,SQLAlchemy
Machine Learning & AI: HuggingFace Transformers,SMOTE,SHAP,Tool Calling,Recall@k,Citation Correctness,Delayed-Label Evaluation,PSI,Statistics
AI/ML Frameworks & Libraries: XGBoost
Generative AI & LLMs: LangChain,LangGraph,OpenAI,GPT-4,Embeddings
Vector Databases & RAG: Retrieval-Augmented Generation,FAISS,BM25,Cross-Encoder Reranking,Metadata Filtering
MLOps: MLflow,Model Registry

Work Experience

Molina Healthcare
United States
AI/ML Engineer
Nov 2024 – Present
Molina Healthcare is a managed-care organization providing government-sponsored health insurance programs; supported the member services division by building LLM-powered policy retrieval and inference systems for clinical and prior authorization use cases.
Tech Stack: FAISS, BM25, OAuth 2.0, Metadata Filtering, Cross-Encoder Reranking, FastAPI, Pydantic, Retry Logic, pytest, GitHub Actions
  • Built a Retrieval-Augmented Generation pipeline using FAISS and BM25 to search 10,000+ healthcare policy documents for the member services division, achieving a Recall@5 of 91% on internal evaluation benchmarks.
  • Developed document-processing and embedding workflows with HuggingFace sentence transformers and OpenAI embeddings to prepare clinical guidelines and prior authorization documents for semantic retrieval.
  • Implemented document-level OAuth 2.0 access controls and metadata filtering so retrieved policy content was restricted to member-tier permissions before reaching the generation layer.
  • Combined dense vector retrieval with BM25 keyword matching and cross-encoder reranking so both semantically relevant and exact-match policy passages could be ranked together before response generation.
  • Exposed LLM inference through FastAPI REST services with Pydantic request and response validation, providing consistent and type-safe interfaces for application requests and generated responses.
  • Added configurable retry logic and fallback model handling to manage transient inference failures, ensuring a reliable response path when the primary OpenAI model could not complete a request.
  • Instrumented request latency and token usage tracking to identify expensive prompt patterns, reducing average token consumption without measurable accuracy loss.
  • Built pytest regression suites covering groundedness, citation correctness, and Recall@k, then integrated them with GitHub Actions CI/CD workflows so retrieval-quality regressions were caught before pipeline changes were deployed.
IDFC First Bank
India
ML Engineer
Dec 2021 – Dec 2023
IDFC First Bank is a retail and commercial bank in India; built and productionized credit risk classification models and batch scoring infrastructure supporting the bank's loan underwriting and regulatory compliance workflows.
Tech Stack: XGBoost, SMOTE, SHAP, MLflow, AWS SageMaker, PostgreSQL, SQLAlchemy, Azure DevOps, PSI, KS Statistics, Delayed-Label Evaluation
  • Developed credit risk classification models using XGBoost on 500K+ loan application records, applying SMOTE to correct class imbalance and time-aware train/validation splits to prevent leakage; the final model achieved a PR-AUC of 0.87 on held-out test data.
  • Engineered applicant-level features from income-to-debt ratios, payment-history aggregations, and bureau-derived credit metrics, reviewing each attribute for leakage risk before including it in the training workflow.
  • Calibrated model probability scores with Platt scaling and tuned decision thresholds to maximize recall within an 8% false-positive budget defined by the bank's credit policy, increasing approval coverage by 12%.
  • Productionized batch scoring workflows on AWS SageMaker with scheduled daily runs, persisting prediction outputs to PostgreSQL via SQLAlchemy ORM and exposing results to downstream risk reporting systems.
  • Registered all model artifacts, training configurations, and evaluation metrics in MLflow Model Registry, enabling one-click rollback and full reproducibility across every promoted model version.
  • Applied SHAP TreeExplainer to generate per-applicant feature contribution reports for regulatory audit readiness, reducing manual explanation preparation time.
  • Monitored production model health through feature drift tests using PSI and KS statistics, triggering automated Azure DevOps retraining pipelines when drift thresholds were exceeded.
  • Implemented delayed-label evaluation workflows that ingested ground-truth default outcomes arriving 30 days post-prediction, computing rolling precision and recall metrics to detect gradual model degradation before business impact occurred.
Netcracker Technology
India
Python Developer
Nov 2020 – Dec 2021
Netcracker Technology provides OSS/BSS software solutions to telecommunications operators globally; built backend microservices and data processing pipelines supporting telecom provisioning and network configuration workflows.
Tech Stack: FastAPI, Flask, Pydantic, REST APIs, PostgreSQL, SQLAlchemy, Retry Logic, Timeout Handling, Docker, Azure DevOps, CI/CD Pipelines, pytest, pandas, NumPy
  • Developed modular FastAPI and Flask microservices for Netcracker's OSS/BSS platform, exposing REST endpoints consumed by telecom provisioning systems and handling 50,000+ daily API requests with structured Pydantic response schemas.
  • Designed PostgreSQL database schemas using SQLAlchemy ORM with normalized table structures, indexed queries, and migration scripts, reducing average query response time by 28% on high-volume provisioning transaction tables.
  • Integrated third-party telecom vendor APIs with configurable retry logic, exponential backoff, and timeout handling, stabilizing connectivity to upstream systems that previously caused cascading downstream failures.
  • Containerized all services with Docker and established Azure DevOps CI/CD pipelines for automated build, test, and deployment, cutting release cycle time from two weeks to three days.
  • Authored pytest unit and integration test suites covering core API contracts, database interactions, and error-handling scenarios, raising code coverage across the service layer.
  • Processed large-scale telecom network configuration data in CSV and JSON formats using pandas and NumPy, transforming raw provisioning records into structured inputs for downstream analytics pipelines.

Projects

Agentic AI System with LangGraph
Tools Used: Python, LangGraph, LangChain, OpenAI GPT-4, Tool Calling, LangSmith
  • Architected a stateful agentic AI system using LangGraph to orchestrate multi-step workflows and LLM-driven task execution with conditional routing and state management.
  • Implemented ReAct-style agent behavior enabling the system to reason over tasks and dynamically select appropriate tools during execution.
  • Integrated short-term and persistent memory modules to maintain context across ongoing interactions and multi-turn sessions.
  • Added human-in-the-loop workflow controls allowing operators to review and modify agent state at defined checkpoints during execution.
  • Instrumented agent executions with LangSmith for tracing state transitions, debugging workflows, and analyzing runtime behavior.
Generative AI RAG Chatbot
Tools Used: Python, LangChain, FAISS, OpenAI GPT-4, HuggingFace Transformers, FastAPI
  • Built a RAG-based conversational chatbot that answers user questions from uploaded documents, grounding responses in retrieved source content rather than relying solely on the LLM's parametric knowledge.
  • Designed a document ingestion pipeline that chunks, embeds, and indexes user-supplied files into a FAISS vector store for efficient semantic retrieval at query time.
  • Implemented a retrieval-augmented generation chain using LangChain to retrieve the top-ranked document passages and pass them as grounded context to the generation model.
  • Exposed the chatbot through a FastAPI service with structured request and response validation, enabling integration with front-end applications.

Powered by Drivetube · Create your own profile at drivetube.ai

Everything on Drivetube

Six products, two of them free forever. Start wherever you are.

Read the docs →
  • Your free digital resume

    at drivetube.ai/in/your-name: one true standard resume with a Hiring Snapshot (visa status, expected salary, notice period, work preference, relocation), an ATS-ready PDF download and a single shareable link. Free forever; interview requests come from verified employers and your contact details stay masked until you accept.

    Documentation
  • Verified jobs, posted in the last 3 days

    verified openings crawled ATS-by-ATS from 160,000+ real company career pages across 50+ ATS platforms. Shows the true posting date from the source ATS — not when a listing was indexed — and deletes every general listing 3 days after it was actually posted. No ghost jobs, no ad-sponsored listings, no staffing reposts, no account needed.

    Documentation
  • Written by senior career writers

    human-written, ATS-optimised resumes by senior career writers, from ₹499.99 / $25.99. Available in every country, written to the destination country's own standard — a US resume, UK CV, German Lebenslauf and Indian resume are genuinely different documents. A paid service, separate from the free Drivetube Profile.

    Documentation
  • Drivetube OneFrom $39.99 / 3 mo

    Daily matched jobs, tailored resumes, or we apply for you

    Drivetube's paid service. Community sends daily Jobsheets of fresh jobs matched to your profile on WhatsApp and email, with the Drivetube Community ($39.99 for 3 months or $69.99 for 6 months; ₹999 or ₹1,899 in India). Jobhunt is one-time application credits with Community included: one credit tailors one resume and its cover letter (100 for $99 with 3 months of Community, 250 for $199 with 6 months). Let Us Apply is an add-on to Jobhunt: a human expert applies 250 times for you on your credits ($0.60 each, $150). Everything lives on one dashboard at drivetube.ai/drivetubeone.

    Documentation
  • For employers — no job postings, no applications

    hiring with no job postings and no applications. Paste your real job description and AI matches it against candidates' true standard resumes, returning ranked candidates with a match %, matched and missing skills and written reasoning. Free tier included; employers pay, candidates never do.

    Documentation
  • Documentation

    every product explained in full, with feature-by-feature comparisons against the job boards, AI apply tools, resume services and hiring platforms people actually use.

  • Playbooks

    tactical job-search guides. Each post is one named tactic with the situation it applies to, the exact steps, a copy-paste message and honest failure modes. Free, no account.

The job board covers the United States, India, United Kingdom, Canada, Europe and Australia, and Resume Writing Services are available in every country.

Pricing is identical on the web, iOS and Android — no app-store markup. Monthly Community membership is purchased in the iOS app only; the website sells annual plans. The Android app has no in-app purchase, so Android members subscribe on drivetube.ai and then sign in to the app with full access. Membership follows your account, not your device.