PRIYANSHU(1)              User Commands              PRIYANSHU(1)

NAME

priyanshu - systems engineer, coffee addict (not in that order)

SYNOPSIS

priyanshu [--role=SDE] [--focus=SCALE]
priyanshu --build [LARGE_SCALE_SYSETEMS] --optimize [EVERYTHING]

DESCRIPTION

Software Development Engineer at Flipkart working on next-gen logging and tracing infrastructure at petabyte scale. Specializes in distributed systems, high-throughput data pipelines, and extreme performance optimization.

Built consensus systems from scratch, slashed P99 latencies by 95%, and processes 200k+ events/sec in production. Writes code in Go, Rust, Java, and C++.

OPTIONS

--languages
C/C++, Rust, Golang, Java, Python, Bash, SQL
--frameworks
Docker, Kubernetes, Apache Pinot, Kafka
--databases
Neo4j, PostgreSQL, Various storage backends
--infrastructure
Linux (Arch/Debian), Git/GitHub, Grafana, Distributed Tracing
--can-explain
Distributed Systems, System Architecture, Backend Development, Performance Engineering
--tools
Custom profilers, Load testing (Locust), Observability stacks

EXPERIENCE

Software Development Engineer 1 - Flipkart

July 2025 - Present | Bengaluru, Karnataka

  • Architected and implemented core components of Flipkart's large-scale logging platform
  • Drove adoption of the above mentioned platform by constantly improving user experience
  • Built high-throughput Kafka pipeline from scratch (200k events/sec)
  • Slashed P99 query latency by 95% via aggressive batching & caching
  • Benchmarked DB backends, schemas, and encodings for production decisions

Software Development Engineer Intern - Flipkart

January 2025 - July 2025 | Bengaluru, Karnataka

  • Built observability infrastructure for Java microservices using custom agent
  • Refactored internal logging framework to resolve conflicts & improve extensibility
  • Performed distributed load testing with Locust to benchmark agent overhead
  • Quantified performance regressions in real-world applications

Research Intern - Indian Institute of Science

May 2024 - July 2024 | Bengaluru, Karnataka

  • Integrated ResNet model into existing pipeline, improving accuracy by 3.7%
  • Fine-tuned embedding extractor on 100 hours of speaker verification data
  • Enhanced speaker diarization pipeline accuracy through model optimization

PROJECTS

Distributed Key-Value Store

  • Engineered distributed consensus in Go from scratch (no libraries)
  • Implemented Raft: Leader Election, Log Replication, fault tolerance
  • Built Log Compaction & Snapshotting for instant crash recovery
  • Solved network partitions with robust RPC timeouts & heartbeats
  • Achieved linearizable consistency across distributed nodes
  • Stack: Go, custom RPC implementation

LiteRAG: High-Performance Java RAG Framework

  • Built vertical RAG framework with custom inverted index & tokenizer
  • Outperformed standard libraries in memory efficiency
  • Implemented delta-encoding & random-access disk layouts for low-latency retrieval
  • Designed BM25 ranking with length normalization
  • Achieved 75% Recall@K on SQUAD datasets (competitive with off-the-shelf solutions)
  • Stack: Java, custom indexing, BM25 algorithm

Kafka Stress Testing Data Pipeline

  • Built from-scratch kafka consumers and producers to validate terabyte-scale throughput
  • Sustained 200k events/sec throughput for extended periods
  • Identified bottlenecks in ingestion layer through systematic testing
  • Stack: Kafka, Java, Apache Pinot

ACHIEVEMENTS

  • Reduced P99 query latency by 95% on production logging platform
  • Built distributed consensus system without external libraries (pure Go)
  • Processed 200k+ events/sec in production Kafka pipeline
  • Improved ML model accuracy by 3.7% through architecture changes
  • Benchmarked multiple storage backends to drive architecture decisions
  • Achieved 75% Recall@K building RAG system from scratch in Java

EDUCATION

B.Tech in Electronics and Communication Engineering
National Institute of Technology Karnataka, Surathkal
December 2021 - May 2025
Focus: Tried learning some Software Engineering, succeeded i guess

METRICS

┌─────────────────────────────────────────────┐
│ Production Stats                            │
├─────────────────────────────────────────────┤
│ Events/sec Processed:   200,000+            │
│ Petabytes Managed:      Multiple            │
│ P99 Latency Reduced:    95%                 │
│ Consensus Nodes:        Linearizable        │
│ Lines of Code:          Increasing          │
│ Coffee Consumed:        Infinite            │
└─────────────────────────────────────────────┘

AVAILABILITY

Currently building petabyte-scale infrastructure at Flipkart.
Always interested in: distributed systems, performance optimization, consensus algorithms, and systems that operate at unreasonable scale.

EXIT STATUS

0
System scaled to petabytes successfully
1
Latency above P99 threshold (unacceptable)
2
Consensus not achieved (network partition detected)
95
P99 latency improvement achieved (success)

SEE ALSO

home(1), blog(1), github(1), linkedin(1), email(1)

BUGS

Occasionally writes Rust when Go would suffice. Otherwise, tries to build cool stuff around core distributed systems.
Report issues to: priyanshu17@outlook.com 

Linux (Arch)             2026-01-18              PRIYANSHU(1)