Amritpal Singh

Software Engineer

Noida, Uttar Pradesh, India4 yrs 10 mos experience

Highly Stable

Key Highlights

Led development of user-focused generative AI solutions.
Architected high-throughput, low-latency platforms.
Published research on DeepFake video detection.

Stackforce AI infers this person is a Backend-heavy Fullstack Engineer specializing in SaaS with a focus on AI and microservices.

Contact

amritpal@pw.live LinkedIn

Skills

Core Skills

Large Language Models (llm)Python (programming Language)GolangRedisMicroservicesNode.js

Other Skills

EmbeddingsFastAPINestJSStrimzi KafkaMqttNext.jsCassandraJavaScriptTypeScriptDockerAmazon S3MongoDBSQLGitHubGo (Programming Language)

About

I recently transitioned to SDE-3 at PhysicsWallah, where I lead the development of user-focused generative AI solutions, including the AI Guru and PW Books. With a backend-first mindset, I partner closely with product and data teams to architect scalable microservices that handle growing loads with reliability. As part of our central engineering team, I contributed to optimizing our MQTT messaging pipeline to support higher throughputs, and introduced automated “big-key” detection scripts that lowered Redis shard memory usage. I also spearheaded the modular decomposition of our legacy test-conducting service into dedicated microservices, Test Microservices to streamline student test-taking workflows, reduce deployment complexity, and enhance maintainability. Previously, I architected Saarthi, a real-time doubt-resolution platform that sustained 200K RPM at low latency through strategic caching and performance profiling. Outside of engineering, I published “DeepFake Video Detection: A Time-Distributed Approach” in SN Computer Science, and I’m passionate about exploring advanced networking in Go and the potential of quantum computation. Let’s connect if you’re tackling generative AI challenges, high-throughput architectures, or simply want to discuss the future of development.

Experience

4 yrs 10 mos

Total Experience

Average Tenure

4 yrs 10 mos

Current Experience

Pw (physicswallah)

4 roles

SDE 3

Apr 2025 – Present · 1 yr 2 mos · Noida, Uttar Pradesh, India

SDE-2 (AI/ML Pod)

Promoted

Sep 2023 – May 2025 · 1 yr 8 mos · Noida, Uttar Pradesh, India

Built backend for AI Books Phase 1, integrating AI-driven summaries and Q&A over text, driving 100 K downloads in the first month and elevating user engagement.
Spearheaded PRD creation: Authoring user stories, data flows, and success metrics to shape the AI chatbot POC roadmap later known as AI Guru.
Architected template & embedding pipelines using Python and LangChain, generating domain-specific prompts for support, academic, and product queries to bootstrap chatbot accuracy.
Introduced LangChain Function Calling, orchestrating AI agents for multi-step reasoning and dynamic data fetches enhancing conversational depth and reliability.
Collaborated with Astra DB team to diagnose and fix performance bottlenecks optimizing query patterns and embeddings issue.

Large Language Models (LLM)EmbeddingsPython (Programming Language)FastAPINestJS

SDE-2 (Platform Team)

Apr 2023 – Aug 2023 · 4 mos · Noida, Uttar Pradesh, India

Conducted POCs on Strimzi Kafka hosting and Golang vs. Node.js performance analysis to guide strategic technology decisions.
Implemented gRPC in the Saarthi doubt-resolution platform, reducing inter-service communication latency by 10–15 ms through targeted traffic analysis.
Developed automated big-key detection scripts for Redis, lowering shard memory usage from 90% to 60% without impacting live traffic stability.
Collaborated on MQTT pipeline enhancements as part of the central engineering team, tuning protocol settings to support increased throughput while maintaining low latency.
Scaled real-time features by load-testing Socket.IO with Artillery—simulating thousands of concurrent connections, measuring response times, and identifying bottlenecks to ensure reliability at scale.

RedisStrimzi KafkaMqttGolang

SDE-1

Jun 2021 – Mar 2023 · 1 yr 9 mos · Noida, Uttar Pradesh, India

Directed the revamp of the Tests module backend redesigning APIs and data models to enhance test-taking experience, delivering revamp PRD in stringent timeline and ensure seamless scaling under peak loads.
Architected and launched Saarthi, a real-time one-to-one doubt-resolution platform, achieving sub-40 ms latency at 200 K RPM through load testing, flame-graph profiling, and strategic caching.
Optimized bulk video processing with parallelized scripts capable of ingesting hundreds of videos concurrently, reducing manual workflow time from 3–4 days to under 3 hours.
Engineered a Python-based result calculation module using machine learning techniques, cutting end-to-end processing from weeks to days and earning founder recognition for operational efficiency.
Implemented Phase 1 of video doubt-solving, enabling in-video question annotations and educator solution uploads to boost student–educator engagement on the platform.

NestJSNode.jsRedisMicroservices