Saurabh Mahapatra

Software engineer with 1+ years of experience building backend systems, ETL pipelines, and REST APIs at two early-stage startups. Shipped production features on real data (10M+ Meta ad records) and automated compliance workflows using Java, Spring Boot, AWS Lambda, and Node.js; comfortable picking up new tools quickly and owning features from design through deployment.

Software Engineer - Marxx.ai - Gurgaon, India

(2025-09)

• Shipped the core REST APIs and data ingestion backend as one of the first engineers; the system now processes 10M+ Meta ad records with under 15 min end-to-end latency.
• Set up async Java workers and a buffered queue between ingestion and downstream consumers, decoupling the pipeline stages and sustaining throughput at 10M+ records per run.
• Added Ad Fatigue Detection that tracks CTR decline over a configurable rolling window and flags underperforming ads, plus a campaign audit endpoint rolling up budget and CPM data — both returned to the React frontend as a single REST call.
• Dropped the failed record rate from 8% to under 1% at 10M+ scale by adding retry logic and routing bad records to a dead-letter queue.

Software Developer Intern - Recur Club - Gurgaon, India

(2024-08 - 2025-08)

• Engineered an ETL pipeline in Java, AWS Lambda (Python), and Node.js to run AML, PEP, and sanctions checks on PAN/DIN records, slashing per-entity verification time by 65% (45 min to 15 min) across 800+ monthly compliance checks.
• Developed a Backend-for-Frontend (BFF) service in Groovy using the Aggregator pattern to consolidate 4 downstream API calls into 1 per page view, trimming frontend network calls by 25% and median page load from 2s to 1.4s.
• Wrote a Python/Flask microservice that pulls data from PostgreSQL and two third-party APIs, generates risk profile reports, and sends alerts via AWS SNS; saved analysts 14 hrs/week of manual review (70% reduction).
• Built a pipeline surfacing PAN, DIN, and corporate records as a linked Related Party graph; analysts trace entity connections from a single dashboard instead of querying multiple systems.

Bachelor of Technology - Computer Science and Engineering - Lovely Professional University (2021 - 2025)