S
Sri Harsha Tenali
About
Detail
Illinois, United States
Résumé
Jobs
verified_user
0% verified
Software Engineer (Contract)BrexJan 2025 - Current (1 year 7 months)Designed and owned backend modules for an enterprise spend management platform serving 35,000+ business customers across expense reporting, ERP integrations, and real-time policy enforcement workflows • Architected RESTful and GraphQL APIs for the expense reporting and GL coding module using FastAPI, supporting multi-currency transaction processing across 150+ enterprise clients • Owned the real-time spend policy rule-evaluation engine, designing configurable enforcement logic that reduced policy violation escape rate by 14–19% across card transaction workflows • Designed ERP integration APIs syncing with QuickBooks and NetSuite using PostgreSQL on AWS RDS, achieving 99.97% sync reliability across scheduled and event-triggered reconciliatio- UResearch AssistantUniversity of Illinois SpringfieldJul 2024 - Dec 2024 (6 months)Prototyped a RAG-based pipeline for financial document retrieval and question-answering using LLMs, evaluated against SEC filing datasets as part of graduate research in applied NLP • Prototyped a RAG pipeline using LlamaIndex 0.10.x and ChromaDB to retrieve and answer questions over SEC filings (10-K, 10-Q, 8-K), achieving Recall@K of 0.81–0.85 against a BM25 keyword search baseline • Evaluated embedding strategies using text-embedding-ada-002 across the FinanceBench dataset (~4,500 QA pairs), benchmarking Faithfulness scores of 0.86–0.92 and Answer Relevancy of 0.78–0.84 • Investigated LLM fallback behavior by integrating gpt-3.5-turbo as primary and Phi-3-mini as local fallback, documenting latency and accuracy tradeoffs across 150+ test
- Software EngineersliceJan 2022 - Nov 2023 (1 year 11 months)Built and maintained backend microservices for a digital lending platform serving 5M–12M registered users across real-time loan origination, repayment, and fraud detection workflows • Developed the repayment scheduler and EMI calculation engine in Python, reducing failed repayment processing events by 18–23% through retry logic and idempotency handling across 200,000+ monthly card transactions • Built a rules-based fraud detection engine evaluating 100+ credit scoring variables per application, achieving 91–94% flagging precision across the real-time loan origination pipeline • Implemented event-driven onboarding workflows using AWS Step Functions and MongoDB Atlas Change Streams, with Kafka event streaming patterns reducing new-user decisi
Education
verified_user
0% verified
- UMaster of Science in Computer ScienceUniversity of Illinois SpringfieldGPA: 3.94/4
- MBachelor of Science in Computer ScienceMRIETGPA: 7/10
Projects (professional or personal)
verified_user
0% verified
- SSecuring Pharmaceutical Data Using Homomorphic Encryption• Designed a two-layer encryption system using AES-CTR encryption (pyaes + PBKDF2 key derivation) and Paillier homomorphic encryption for numeric quantities, ensuring all data remained encrypted at rest in MySQL • Implemented additive homomorphic operations on encrypted medicine quantities using the Paillier cryptosystem, enabling inventory updates directly on ciphertext without decryption • Built a role-based access control system in Django separating Employee and Manager permissions, with 256-bit Paillier keypair generation per medicine entry and AES-encrypted credential storage
- gService Booking Platformgithub.comtenalisriharshaservice_booking_platform• Built a full-stack service booking platform using Django REST Framework and React.js with JWT authentication, role-based access control across client, provider, and admin roles • Designed RESTful API endpoints for booking slot management, provider availability configuration, and service search with date and name filtering, tested end-to-end via Postman • Architected a modular Django backend with separated concerns across accounts, booking, payments, and notifications apps, with SQLite in development and PostgreSQL-ready configuration for production
- gAi Assistant for Appointment Schedulinggithub.comtenalisriharshaai-assistant-app• Architected a Flask REST API backend with a React.js + TypeScript frontend supporting natural language appointment scheduling via Grok API integration, with LLM-independent fast-path handlers for 12+ intent types processing common intents locally without API calls • Built a conflict-aware recurring event engine in Python supporting weekly, interval, and count-bounded recurrence patterns with cross-midnight window handling, bulk conflict detection, and skipped-conflict surfacing • Integrated Vosk-based local voice transcription as a speech input layer feeding the same NL processing pipeline, with a reminder polling system supporting toast notifications and snooze state management persisted via SQLite