Professional Summary
Backend Engineer with nearly 3 years on large-scale, event-driven systems in Java/Spring Boot, Kafka, and PySpark/Databricks, building the recommendation and feed infrastructure behind a platform serving 100M+ users. I specialize in changing live systems safely — dual-write migrations, parity testing before cutover, staged rollout and rollback.
Pursuing an M.Tech in AI & ML (BITS Pilani) and moving toward ML platform and inference infrastructure — bringing production reliability and migration discipline to how models get served. Hands-on ML work to date is in personal projects: RAG pipelines, LLM-powered services, and an end-to-end MLOps pipeline.
Technical Skills
Languages
Java, Python, SQL, JavaScript
Backend & Streaming
Spring Boot, Kafka, Spark Structured Streaming, Protobuf, gRPC, REST APIs
Data Platform
Databricks, Apache Spark, PySpark, dbt, Unity Catalog, BigQuery, ClickHouse, ETL Pipelines
Infrastructure & Delivery
Kubernetes, Helm, Argo CD, KEDA, Terraform, Docker, GCP (GKE, GAR), GitOps, CI/CD
Observability & Caching
Prometheus, Grafana, Loki, metrics & alerting, Redis, Caffeine
Practices
Dual-write migration, parity testing, staged rollout & rollback, contract evolution, least-privilege access
ML & AI (projects & coursework)
PyTorch, HuggingFace Transformers, RAG, Vector Search (ChromaDB), MLflow, FastAPI
Education
M.Tech. Artificial Intelligence & Machine Learning (April 2026 - Present)
BITS Pilani — Work Integrated Learning Programme
Bachelor of Technology in Computer Science (CGPA: 9.10 / 10.00) 🥈 Silver Medalist
Rajasthan Technical University — Aug 2019 to July 2023
Experience
Sept 2023 – Present
Software Development Engineer (SDE-1), Backend
Glance, InMobi — Bengaluru, IN
- Built the L0 content-enrichment service in Java/Spring Boot — a Kafka consumer that joins recommendation-planner output with content APIs and emits Protobuf feed messages — on a personalization platform serving 100M+ users
- Migrated enrichment off a PySpark/Databricks path with zero downtime: dual-write to a shadow topic, a 12-row parity specification with fixtures as the correctness gate, then a staged ramp from 1% to 100% of user buckets with a one-line rollback on standby
- Replaced a monolithic enricher with an ordered, failure-isolated pipeline of 12 units across 7 reviewed phases, so a single failing unit degrades one widget instead of the feed, and emits failure telemetry
- Hardened Spark Structured Streaming jobs on Databricks — auto-retry, checkpointing before Kafka writes, delivery semantics, elastic disk — and traced silent candidate loss across a service boundary using on-wire Protobuf decode diagnostics, without altering the production write path
- Owned the production path across 7 repositories: Kafka topics and ACLs, Terraform and Unity Catalog least-privilege grants, dbt sources with freshness and duplicate checks, Argo CD/Helm/KEDA autoscaling, and Prometheus/Loki observability
- Earlier: built a gRPC-based dynamic configuration service with semantic versioning for backward-compatible rollouts, a GDPR-compliance API with region-based access controls, and a release-automation platform (Python/Flask) on GKE with Helm and CI/CD
Tech: Java, Spring Boot, Kafka, Protobuf, gRPC, Spark Structured Streaming, Databricks, PySpark, dbt, Unity Catalog, BigQuery, ClickHouse, Redis, Caffeine, Kubernetes, Helm, Argo CD, KEDA, Terraform, GCP (GKE, GAR), Prometheus, Grafana, Loki
May 2022 – July 2022
Machine Learning Engineering Intern
Celebal Technologies — Jaipur, IN
- Built a video recommendation system on user-interaction data, from feature preparation through offline evaluation
Tech: Python, Machine Learning, Recommendation Systems
Projects
Content Intelligence Pipeline
AI / MLOps — github.com/PranavSagar/content-intel-pipeline
- News-classification service: fine-tuned DistilBERT to 94.64% on AG News, served over FastAPI with Kafka streaming, MLflow experiment tracking, and Evidently drift checks; deployed on HuggingFace Spaces
Tech: Python, PyTorch, HuggingFace, FastAPI, Kafka, MLflow, Evidently, Docker
RAG Football Wiki
AI / RAG — github.com/PranavSagar/rag-football-wiki
- Retrieval-augmented Q&A service: sentence-transformer embeddings in a ChromaDB vector store, retrieved as context for an LLM (Claude) to generate grounded answers; served via FastAPI
Tech: Python, LLM (Claude), sentence-transformers, ChromaDB, FastAPI
Achievements
🏆 Avengers Award at Glance (InMobi)
🥈 Silver Medalist - Rajasthan Technical University
📜 Microsoft Certified - Azure AI & Data Fundamentals
💻 LeetCode Rating 1577 (Top 25%)
🥈 Smart India Hackathon 2022 Runner-Up