Vineet Malewar vintu2001

Hi, I'm Vineet Malewar 👋

Software Engineer · Distributed Systems & Database Internals

San Francisco Bay Area, CA

🚀 Software Engineer specializing in distributed systems, database internals, and cloud-native storage abstractions.
🛠️ Deeply interested in open lakehouse architectures (Apache Iceberg), decoupled storage-compute environments, and consensus protocols (Raft).
💻 Experienced in building high-concurrency, low-latency system components in C++20 and Java.
🎓 Currently pursuing my M.S. in Computer Software Engineering at San José State University (Expected May 2027).

Tech Stack: C++20, gRPC, PostgreSQL, AWS EC2, CMake

Architected a high-concurrency metadata control plane in C++20 for open table formats (Apache Iceberg) storing columnar data (Parquet).
Implemented MVCC-based versioning in the storage layer to expose consistent snapshot reads over gRPC during concurrent DDL operations.
Optimized for decoupled storage-compute environments, achieving sub-10ms metadata retrieval for 1M+ active data partitions.

Tech Stack: Java 17, Raft Consensus, GKE, Docker, GCS

Built a production-grade distributed object storage system, orchestrating cluster metadata and chunk placement decisions via the Raft consensus algorithm.
Implemented a consistent hashing ring for request routing with quorum-based replication (RF=3) for cross-node data durability.
Containerized infrastructure deployed via GKE StatefulSets, persisting chunk data to dedicated GCP buckets.

Tech Stack: Java 17, Spring Boot, Azure Event Hubs, SQL Partitioning

Developed a distributed ledger supporting concurrent multi-asset billing transactions with strict ACID compliance.
Leveraged Azure Event Hubs for high-throughput event ingestion, maintaining an immutable audit trail with 99.99% durability.
Achieved sub-50ms validation latency for atomic transactions via an optimized SQL partitioning strategy.

Tech Stack: Python, FastAPI, Docker, LlamaIndex, Ollama, AST

Containerized agent unifying siloed developer data → estimated 75% faster time-to-information.
Code→Graph pipeline via Python AST + networkx, parsing 15k+ LOC (requests library) into a knowledge graph of 290+ symbols.
Agentic RAG with Llama 3 (local) + custom tools for graph querying. Real-time interface via Discord Bot + REST backend → answers in <5s.

Tech Stack: Java, Spring Boot, Kafka, MongoDB, REST

High-volume ingestion from multiple sources into centralized MongoDB using Kafka + Spring Boot.
Full-text search & flexible field filters via REST for precise, fast retrieval.

Core Domains: Distributed Consensus (Raft), Storage-Compute Decoupling, Catalog Management, Concurrency Control, Multi-threading, Agentic RAG.
Data & Formats: Apache Iceberg, Apache Parquet, Apache Flink, MVCC.

"Building reliable distributed systems at scale."