Career
Professional
Experience
Measurable impact across strategy, engineering, and analytics.
📍 Hyderabad, Telangana (Remote)
- Built and deployed Deep Research RAG, a production-oriented research document intelligence system processing 444 pages and 3,439 chunks from the Infosys FY2025-26 annual report and AGM transcript using PyMuPDF, recursive chunking, MiniLM embeddings, and ChromaDB.
- Engineered a multi-stage retrieval pipeline combining dense retrieval, BM25 sparse retrieval, Reciprocal Rank Fusion (RRF), HyDE query expansion, and cross-encoder reranking, with page-level deduplication and source and page citations for grounded generation.
- Diagnosed and improved retrieval failures through controlled experiments, including dense-retrieval false positives and hybrid-ranking regressions. HyDE plus BM25 improved the tested North America revenue query from MRR 0.00 with dense retrieval to 1.00, demonstrating empirical retrieval debugging rather than framework-only implementation.
- Evaluated the RAG pipeline with 13 ground-truth QA pairs using RAGAS (Context Precision 0.991, Context Recall 1.000) and deployed the system as a Dockerised FastAPI application on Railway with persistent ChromaDB, cached Hugging Face models, and separate demo and API interfaces.
- Built Learning Sprint Planner Agent: a LangChain ReAct agent (create_agent) using custom tools for course lookup and completion-time calculation, generating structured learning plans from topic, skill level, completed hours, and daily study time.
- Separated deterministic logic from LLM reasoning: Pydantic validation and calculations run in Python; the LLM handles tool selection and plan generation. Also built an LCEL Explain mode (ChatPromptTemplate, StrOutputParser) with Streamlit.
- Built lightweight AI automation workflows including an n8n-based daily intelligence digest and an LLM-powered pre-call research briefing. Explored MCP-based tool augmentation and vector database architectures using ChromaDB and Pinecone.
📍 Bengaluru, Karnataka
- Engineered AI-driven research workflows leveraging Claude, ChatGPT, and Perplexity to extract market sizing and KPI data from annual reports, embedding source-verified figures into Excel and reducing document processing time by 50%.
- Spearheaded a Vendor Due Diligence engagement in natural animal healthcare by defining project scope, conducting requirements analysis, benchmarking five competitors through expert calls, and delivering scenario-based market insights to C-suite stakeholders.
- Streamlined financial benchmarking across six global competitors during an IPO readiness project by creating an automated Excel model that normalised currencies and fiscal-year differences, cutting review time by 80% and ensuring on-time delivery.
- Mentored and guided 1–2 analysts per project across multiple workstreams, providing structured training and checkpoint reviews that restored delivery quality without adding to personal execution load.
- Delivered end-to-end strategy and analytics solutions across financial services, technology, and healthcare, applying quantitative frameworks to support investment theses, opportunity sizing, and business case development for senior leadership.
📍 Hyderabad, Telangana
- Contributed to the end-to-end migration of a client banking portal from AngularJS to Angular 11, developing reusable UI components and a custom Angular widget across HTML5, CSS3, and JavaScript/TypeScript, with Git for version control and SQL/PostgreSQL for data validation, across 2+ years of front-end delivery in an enterprise BFSI environment.
- Identified and resolved a performance issue on a banking portal: traced a 3 to 4 second page load delay to repeated database calls through root-cause analysis of system data, and presented documented findings and resolution recommendations to the project team.
- Built and shipped features every sprint using JavaScript, TypeScript, HTML5, CSS3, and Angular on the frontend, with Java on the backend.
- Handled regression testing and API testing with Postman as a standard part of the delivery cycle.
- Coordinated cross-functional Agile/Scrum teams spanning developers, QA, and project managers, managing JIRA backlogs, tracking sprint milestones, and maintaining structured documentation in Confluence.
📍 Hyderabad, Telangana
- Built and evaluated linear and logistic regression models as part of the data science team workflow, applying core model evaluation techniques including Precision, Recall, F1, RMSE, and R-squared to assess performance.
- Gathered analytical requirements from the data science team, designed and built Power BI dashboards and Excel-based reports tracking key business KPIs, and prepared structured documentation for data processes and reporting outputs.
- Extracted, validated, and cleaned datasets using Python and SQL, ensuring delivery of accurate, analysis-ready data for business reporting and decision-making.
📍 Alwar, Rajasthan
- Conducted field-based stakeholder interviews and requirement assessments for social programme delivery, synthesising findings into structured reports and actionable recommendations for programme leadership.