Prachi Suryawanshi
Backend software engineer with 7+ years designing and operating distributed systems at scale, specializing in Go and Java microservices, cloud-native AWS infrastructure, and platform engineering. Led a team of 8 at Telus Digital to a 95% reduction in production incidents while delivering backend infrastructure for LLM fine-tuning and AI-assisted hiring products.
Experience
Lead Software Engineer
Telus Digital
Bengaluru, India
- Led and mentored a team of 8 backend engineers across full-cycle delivery of 20+ features per quarter; drove a 95% reduction in production incidents through architectural governance and code quality standards.
- Architected and operated distributed, event-driven microservices in Go and Java for Fine-Tune Studio — a high-concurrency task-execution platform for training LLMs and generative AI models — backed by PostgreSQL, Redis, and RabbitMQ.
- Implemented fine-grained RBAC across all platform APIs, replacing coarse-grained permissions with least-privilege access controls to harden security at scale.
- Drove a 30% reduction in system latency by identifying and resolving backend performance bottlenecks — introducing database indexes, implementing materialized views for read-heavy workloads, and decomposing a high-latency monolithic API into smaller, focused endpoints. Built platform-wide observability using Sentry, New Relic, and CloudWatch — structured logging, metrics, and alerting — enabling proactive incident detection and response.
- Expanded platform to support multimedia content and multilingual workflows using AWS S3 signed URLs for secure, time-limited access to private media assets.
- Co-led backend development of a greenfield automated candidate assessment platform in Go and PostgreSQL on AWS, eliminating a fully manual evaluation process and scaling to 10,000+ structured Q&A assessments per month — reducing time-to-hire and cost per evaluation.
Software Developer II
Telus Digital
Bengaluru, India
- Designed and built an automated CI/CD pipeline for daily releases in Python using AWS Step Functions and Lambda, converting weekly manual releases across 6 backend services into daily automated deployments — eliminating 3–4 hours of release manager effort per cycle and removing developer PR merge freezes by introducing a dedicated UAT environment.
- Engineered CircleCI build verification via polling-based status checks, automated tag-based deployments, QA approval gates, zero-downtime rollback, secrets management via AWS Secrets Manager, and Slack-integrated failure alerting; evaluated and chose Step Functions over CircleCI orchestration for superior error handling, retry logic, and built-in audit trails.
- Built a company-wide RabbitMQ Dead Letter Queue management tool in Python, enabling on-call engineers to inspect, search, reprocess, or discard thousands of failed messages daily without disrupting live queue processing; enforced idempotency validation before message actions and integrated CloudWatch audit logging to track all on-call interventions.
- Served as informal backend lead for a 2–3 person team, owning architectural decisions, driving feature delivery, and reducing tech debt through query optimizations that improved API response times across core services.
Software Developer
Chegg Inc
Delhi, India
- Consolidated 4 citation product websites from fragmented PHP, Ruby on Rails, and C# codebases onto a unified Node.js/TypeScript backend — architecting a shared BFF layer, GraphQL APIs, and 4–5 reusable TypeScript packages, reducing maintenance overhead and accelerating feature development across all four products.
- Designed and built CircuitBreaker++, a production-grade resilience library in Java implementing circuit breaking, retry policies, and bulkheading across microservices; adopted by 7–8 engineering teams company-wide.
- Built and shipped a real-time chat-based tutoring platform in Spring Boot with WebSocket messaging, serving 10,000+ users in production.
- Remediated a critical security vulnerability by converting publicly exposed S3 buckets to private storage with signed URL access controls, eliminating unauthorized data access across customer-facing assets.
- Wrote a blog on Spring State Machine Vs In-house Solution
Research Assistant
Loughborough University London
London, UK
- Developed a hybrid ANN model that predicts one-step-ahead USD/INR Exchange Rates with 88% accuracy.
- Research paper accepted for IEEE Technically Sponsored Intelligent Systems (InteliSys) conference. Utilized Hybrid approach of Artificial Neural Network, Linear Optimization, and Sliding Window techniques.
Education
Indian Institute of Technology, (IITR)
B.Tech
Roorkee, India
Skills
GoTypeScriptJavaPythonSQLAWS (Step Functions, Lambda, S3, Secrets Manager)DockerCI/CD (CircleCI)Distributed SystemsRESTGraphQLRedisRabbitMQObservabilityMonitoringAlertingGitGinDjangoSpringNode.jsPostgreSQLNoSQL (MongoDB)
Projects
Automatic Cataloguing of Documents using Artificial Intelligence
Automated book cataloguing process using Optical character recognition and Natural Language Processing. Developed an application that extracts text from images and classifies it into various fields such as author and publisher.
Exchange Rate Prediction
Utilized ARIMA for linear modelling of time series and Backward Regression Analysis for variable selection. Incorporated the results into Neural Network to predict One-step-ahead daily USD/INR.