Samarth Mahendra
I obsess over the parts of software that are invisible — storage engines, distributed runtimes, metadata indexing. Currently at Northeastern, Boston, building things from scratch that most people use off-the-shelf.
— easy · — medium · — hard
— active days in the past year
About Me
📍 Currently based in Boston, MA
Passionate about backend systems, distributed architecture, and AI.
I’m a Backend & Systems Engineer with a strong focus on distributed systems, storage, and data-intensive infrastructure. I enjoy working close to the metal—reasoning about performance, correctness, and failure modes in production systems.
Previously at Draup ($20M+ ARR AI platform), I worked on core backend and data-platform systems including API design, database migrations, async pipelines, and query optimization. I led large-scale API migrations and designed dynamic query frameworks powering production workloads.
Outside of work, I build systems to deeply understand infrastructure fundamentals: a custom on-disk storage engine (B-Trees, WAL, crash recovery), a Ray-style distributed runtime, and hierarchical metadata pruning for lakehouse-style file selection and query planning.
Education
Northeastern University
MS in Computer Science (Jan 2024 – Dec 2025)
Coursework: Programming Design Paradigms, Algorithms, NLP/ML, Database Systems, Mobile Development, Computer Systems, Software Engineering
Dayananda Sagar College of Engineering
BE in Computer Science (Aug 2018 – Jul 2022)
🤝 Boston Community
Regular attendee — an awesome community of Boston developers who meet over code and coffee. Great space for learning, side projects, and meeting fellow builders.
Active participant in Boston-area AI & ML meetup community — engaging with the latest in LLMs, agents, and real-world AI applications across the city.
Skills
🟦 Backend Engineering
Python
Java
C/C++
Rust
TypeScript
Django
REST
Node.js
FastAPI
REST APIs
WebSockets
Async I/O
Concurrency
Caching
(Redis)
Rate Limiting
API Optimization
Message
Queues
gRPC & Protobuf
GraphQL
PyTest
JUnit
TDD
🟪 System Design & High-Performance
Apache Kafka
🟩 Databases & Infra
PostgreSQL
MongoDB
Redis
Elasticsearch
DB Internals
MVCC
Query Planning
Partitioning
Storage Engines
Data Pipelines
Observability
🟧 AI & Automation Systems
OpenAI
GPT-4o
Google
Gemini
Function-Calling Agents
LLM Orchestration
RAG/Embeddings
Automation
Async Agents
Voice
Systems
🟨 Cloud & DevOps
AWS
Terraform
Docker
Kubernetes
CI/CD
IaC
OpenTelemetry
Distributed Tracing
Deployment
Autoscaling
Service
Mesh
Experience
My professional journey.
Associate Software Development Engineer
Draup · Bengaluru, India
- Draup: $20M+ ARR AI-Driven Sales/Talent Intelligence Platform trusted by 5 of the Fortune 10 for actionable market intelligence.
- AI & DevTools: Built an internal AI agent (AutoGPT-based) similar to Claude Code to automate code fixes via SonarQube reports, handling complex indentation and syntax correction.
- System Design: Architected a dynamic query framework and boolean filtering engine (nested conditions), reducing new feature development time by 80%.
- Scale: Led migration of 100+ APIs from PostgreSQL to Elasticsearch, achieving 5x faster aggregations and 400% query optimization.
- Infrastructure: Implemented subscription-based access control and reduced platform downtime from 4% to 1% using Datadog & AWS CloudWatch.
- Core Features: Engineered critical modules including Talent Metrics, Account Insights, and Digital Tech Stacks, delivering actionable business intelligence.
Backend Engineering Intern
Draup · Bengaluru, India
- Optimization: Cut image load times by 70% using Redis caching and delivered a technical talk on caching strategies to the engineering team.
- Observability: Reduced incident response time by 30% by setting up Datadog monitoring pipelines.
Research
Dayananda Sagar College of Engineering
- Patent (IN 202341086278): Co-inventor of "Myocardium Wall Motion and Wall Thickness Mapping" (Filed Aug 2024), developing visual maps to correlate wall motion, thickness, and fibrosis from cine-series MRI scans.
- Algorithm Design: Developed custom algorithms to calculate wall thickness and interpolate indistinct pixels using angular intersection methods.
- Optimization: Enhanced performance by 60x using GPU acceleration (CuPy) and multiprocessing (ThreadPoolExecutor) to process high-res DICOM images.
- Impact: Identified reduced wall motion in heart disease patients and integrated a fibrosis map for comprehensive cardiac analysis.
Writing
Deep dives on systems, storage, and infrastructure
Rethinking Metadata Indexing in Analytical Data Systems
How modern lakehouses handle file selection and query planning at scale — and why naive approaches fall apart when you have millions of files across partitions.
Modern Storage Engines
A ground-up look at how storage engines are built — B-Trees, LSM trees, Write-Ahead Logs, and the design decisions that separate fast databases from slow ones.
Featured Projects
What I've been building.
ButterDB
About ButterDB is now a multi-threaded, persistent Key-Value Store built on a B-Tree structure. It features a custom paging engine, buffer pool, write-ahead logging (WAL) for durability, and fine-grained locking for concurrency.
View CodeLakehouse Metadata Pruning
Hierarchical metadata index enabling efficient file pruning for range queries over large lakehouse datasets.
View CodeReal-Time AI Voice Assistant
Full-stack AI assistant handling two-way voice calls via Twilio & GPT-4o with sub-500ms latency.
View CodeJobStats Analytics
Distributed scraping pipeline capable of tracking 50K+ postings/day with 99.9% uptime.
View CodeIntelligent Agent System
Multi-LLM framework using GPT-4 & Gemini with Discord integration.
View CodeSecret Hitler AI
AI-powered game with 5+ GPT-5 agents that bluff, deduce, and strategize. Features personality-driven table talk.
View CodeStackOverflow Clone
Full-stack Q&A platform with voting, commenting, and reputation systems using React, Node.js, and MongoDB.
View CodeStock Portfolio Manager
Java-based application for managing stock portfolios with MVC architecture and 100+ JUnit tests.
View CodeLet's Work Together
I'm currently open to new opportunities. Whether you have a question or just want to say hi, I'll try my best to get back to you!
Boston, MA, USA | +1 (857) 707-1671