Shruthi Kodati
Software Engineer & Computer Science Graduate Student
Software Engineer and CS graduate student (MS, UIC - May 2026) with hands-on experience designing, building, and shipping secure, scalable full-stack systems, from React frontends and RESTful API design to distributed backend architecture and cloud deployment. Delivered 40% reduction in manual effort and 35% faster response times on production systems real users depend on daily.

Statistical parity disparity eliminated as published in ACM
Model accuracy maintained while debiasing targeted ads
Manual clinical dietary processing effort reduced via greenfield web app design
Response time reduction across 12+ REST API endpoints under production load
Chicago crash records analyzed for predictive safety modeling
Encrypted file transfer size supported across zero-knowledge platform
Continuous production uptime sustained across 6+ months
Lines of cryptographic node logic built for custom messaging
Memory footprint reduction of geospatial analytics parsing
Multi-Modal Table Recall@5 retrieved via hybrid FAISS + BM25
Publications & Research
ACM Peer-Reviewed Publication
Peer-ReviewedFairness in Targeted Advertisements
Co-authored an 11-page paper contributing to responsible AI and algorithmic fairness in production advertising systems. Combines Meta's Ad Library, FairJobs benchmark datasets, and synthetic bias mitigation strategies to achieve significant statistical parity improvements while maintaining near-perfect model performance. Includes full ablation studies across 3 datasets with mathematical formalization and statistical significance testing.
Experience
AI Engineer
- • Built and deployed AI pipelines using LLMs, RAG, and MCP to support research workflows for 5+ academic teams.
- • Developed Python backend services and REST APIs that handle 1M+ requests per month.
- • Optimized SQL queries and vector search to reduce bottlenecks and improve accuracy under heavy load.
Software Engineer - Graduate Research Assistant
- • Spearheaded greenfield development and cloud deployment of a live production web application using Python, Flask, RESTful APIs, HTML5, CSS3, and JavaScript on Render via Git/GitHub CI/CD, automating end-to-end clinical dietary workflows for 3+ active nursing research teams and eliminating 40% of manual processing effort across 500+ structured dietary records per workflow cycle.
- • Re-architected 12+ RESTful API endpoints by restructuring SQL query logic, introducing indexed relational schemas, and optimizing data pipeline throughput, achieving a 35% reduction in average API response time and hardening reliability under concurrent production load serving 10+ daily active users.
- • Engineered a responsive, accessible frontend with DOM-driven dynamic rendering, client-side form validation, and real-time user feedback across 8+ core workflow views, empowering non-technical clinical staff to confidently interact with complex structured data without engineering support.
- • Designed and implemented secure authentication and session management flows, multi-layer server-side input validation, and structured CRUD operations covering 10K+ managed database records, building the complete application stack to production engineering standards with defensible, maintainable architecture.
- • Owned the full CI/CD delivery lifecycle including pipeline configuration, zero-downtime version releases, Git branching strategy, and performance profiling, sustaining 100% production uptime across 6+ months of continuous active development.
- • Developed 6+ dynamic REST API endpoints for real-time data querying, structured aggregation, and clinical report generation, enabling nursing researchers to programmatically filter, access, and export structured dietary datasets directly from the live production system.
- • Partnered with nursing researchers and faculty across 10+ requirement sessions to elicit specifications, validate features, and drive rapid iteration cycles, delivering version-controlled zero-downtime production releases that 15+ active clinical users depend on daily.
- • Conducted systematic debugging, code reviews, and production performance profiling across the full application stack, proactively resolving 40+ critical bottlenecks and documenting all technical decisions to maintain engineering standards throughout the complete project lifecycle.
Software Engineer, Intern
- • Engineered and deployed a full-stack research portal using React.js, Node.js, MongoDB, JavaScript, and Tailwind CSS, digitizing administrative workflows to eliminate manual tracking for 100+ faculty members.
- • Architected responsive user interfaces, modular RESTful APIs, and database-driven backend services, streamlining document submissions, publication indexing, and real-time project milestones.
- • Collaborated across 9 academic departments to analyze complex business processes, gather structural requirements, and translate them into maintainable, scalable software solutions.
- • Streamlined cross-departmental communications and research accessibility, reducing administrative processing latency by 40% and enhancing overall operational efficiency.
- • Optimized application performance and reliability by auditing database schemas, refining API endpoints, and implementing strict client-side and server-side data validation schemas.
- • Executed the end-to-end Software Development Life Cycle (SDLC), troubleshooting and resolving 30+ critical frontend/backend bugs during local test sprints and staging deployment phases.
Salesforce Developer Intern
- • Developed and enhanced CRM features using Apex and Lightning Web Components (LWC), applying object-oriented programming principles to automate business workflows that reduced manual processing steps by 30%.
- • Integrated external systems using REST APIs, JSON, and OAuth 2.0, improving secure data exchange and increasing cross-system data accuracy by 25%.
- • Debugged, tested, and deployed enhancements using Salesforce CLI, Git, and VS Code, improving feature release turnaround time by 20% through structured testing and version control.
Projects
Cipher-Signal Protocol Encrypted Messenger
- • Implemented the complete Signal Protocol stack (X3DH key agreement + Double Ratchet algorithm) from scratch in Node.js and Socket.io, producing a zero-knowledge platform supporting 25MB encrypted file transfers across 12 file types with forward secrecy and post-compromise recovery enforced at the protocol layer.
- • Hardened security with Ed25519 SPK signature verification, 256-bit session tokens, rate limiting, and replay attack prevention, eliminating path traversal vulnerabilities via UUID-namespaced storage and guaranteeing exactly-once message delivery via an ACK-based persistent offline queue with deduplication.
Multi-Modal RAG for SEC 10-K Financial Document Analysis
- • Architected a two-stage RAG pipeline ingesting 150 SEC 10-K filings (8GB, 45K text chunks, 12K tables) with hybrid FAISS+BM25 retrieval and table row-sentence representation, achieving 85% Table Recall@5 and a 12-point Exact Match improvement over text-only baselines.
- • Deployed the production retrieval system (LLaMA 3.2, sentence-transformers, 8+ Flask API endpoints) with intelligent query routing and citation generation, attaining 96% faithfulness and 4.2s median latency across FinQA, DocFinQA, and FinDER benchmarks with full ablation studies across 3 evaluation datasets.
SF Property Tax Analysis & Gentrification Risk Assessment
- • Constructed an end-to-end analytics pipeline processing 3.7M property tax records (6GB GeoJSON) across 93 San Francisco neighborhoods, implementing memory-optimized ETL with categorical encoding and streaming parsing that slashed data footprint by 87% (5.7GB to 733MB).
- • Surfaced 59.6% median property appreciation (2015 to 2023) via 15+ interactive Altair visualizations identifying high-risk displacement zones and quantifying COVID-19 housing market impact for urban planners and policymakers.
Systematic Benchmarking of Multiple Sequence Alignment Algorithms
- • Evaluated and benchmarked computational efficiency and alignment accuracy of five prominent Multiple Sequence Alignment (MSA) algorithms (Clustal Omega, MUSCLE, MAFFT, Kalign, and T-Coffee) using reference alignments from BAliBASE 3.0.
- • Analyzed resource usage (execution time and memory footprint) across 150+ protein sequence datasets, correlating algorithm complexity with sequence length and divergence levels.
Chicago Traffic Crashes: Predictive Safety Analytics
- • Analyzed 938K crash records with 48 features, building ensemble models (Random Forest, XGBoost, Neural Networks) achieving 0.597 RMSE for injury prediction.
- • Discovered 90% pedestrian injury rate and identified highest-risk corridors.
Fairness in Targeted Advertisements
- • Built comprehensive fairness evaluation framework addressing bias in targeted advertising by combining Meta's Ad Library (66K employment ads), FairJobs benchmark dataset (1M+ rows), and custom synthetic data.
- • Achieved 93.4% reduction in statistical parity difference while maintaining 99.2% accuracy, and improved Disparate Impact Ratio from non-compliant 0.70 to near-compliant 0.98.
Technical Arsenal
Languages
Databases
Frontend & Backend
Data & ML
Security
Cloud & DevOps
Visualization
Education
Master of Science in Computer Science (GPA: 3.6/4.0)
University of Illinois Chicago
Aug 2024 – May 2026
Bachelor of Technology in Computer Science (GPA: 8.08/10)
CVR College of Engineering
Aug 2020 – May 2024
Let's Connect
I'm currently open to new roles and exciting collaborations. Whether you have a question, an opportunity in mind, or just want to connect, my inbox is always open. I'd love to hear from you!