Oct 2024 – Dec 2025 · 1 yr 2 mos · Bangalore Urban, Karnataka, India · Hybrid

• Building the AI layer of Wrike's developer experience, from RAG systems and MCP servers to agentic code review workflows and AI-powered developer tooling.
• Designed and deployed Wrike internal MCP server as a centralised context management system for AI-powered IDEs, reducing frontend team manual work by 90% during tech migration task across 100+ mono repos.
• Contributed to architecture and development of the AI Portal, a Vertex AI-powered internal RAG system handling 10K+ daily queries and reducing knowledge discovery time by 40%.
• Architected AI-powered developer tools for automated code review, coding convention suggestions, and context-aware code generation.
• Developed AI-driven prototype generation tool leveraging Wrike's internal design system, reducing design-to-development prototype turnaround by almost 50%.
• Tech Stack: Python, Typescript, LLM, MCP, Vertex AI

Oct 2024 – Aug 2025 · 10 mos · San Francisco Bay Area · Remote

• Worked with AI researchers at Meta and Gemini to train LLaMA 4 using RLHF, contributing to code generation quality improvements and error reduction across production model variants.
• Engineered and optimised 500+ prompts for LLaMA 4 and Gemini, achieving ~35% improvement in code generation accuracy and 20% reduction in hallucination rates on complex debugging benchmarks.
• Built and ran LLM evaluation pipelines using RLHF-based human preference signals and automated metrics (BLEU, pass@k, CodeBLEU), iterating across 1000+ training examples to systematically close model performance gaps.
• Applied RLHF and transfer learning to fine-tune LLMs on domain-specific code datasets, improving coherence, instruction-following, and contextual understanding for code review, debugging, and completion tasks.

Jan 2024 – Oct 2024 · 9 mos

• Designed scalable microservices-based backend supporting 100K+ concurrent users with <100ms latency using AWS Lambda, Redis, and Apache Kafka directly applicable to low-latency LLM inference and streaming AI pipelines.
• Built GenAI features end-to-end: GPT-4 powered event generation (60% faster setup), automated session summarisation saving 200+ hours/month, real-time translation and transcription via Whisper API, and semantic user matching for networking rooms.
• Drove AI adoption for the engineering team, evaluating and rolling out AI-assisted development tools including Cursor, GitHub Copilot, and Gemini, improving developer productivity and reducing implementation turnaround.
• Mentored 2 engineers, led architecture reviews, contributed to technical roadmap planning, and drove cross-functional alignment between product and infrastructure teams.
• Introduced engineering best practices: observability, fault-tolerant design, API versioning, CI/CD (GitHub Actions), reducing incident response time and onboarding friction.
• Served on interview panels for engineering hiring, contributing to technical assessment and team growth.

Jan 2022 – Dec 2023 · 1 yr 11 mos

• Built core components of a real-time service framework using AWS Lambda, Kafka, Firebase, React, and Java, enabling low-latency interactions for large-scale live events.
• Drove performance optimisation initiatives across the platform, reducing system response times for high-concurrency event workloads.
• Designed and implemented scalable backend services for critical platform features, contributing to architecture decisions that supported rapid product growth.
• Worked with product managers and designers to translate complex business requirements into technical solutions, ensuring alignment between engineering output and product goals.
• Built CI/CD pipelines using GitHub Actions, improving deployment speed and reliability across the engineering team as part of platform team.
• Improved system reliability and fault tolerance workflow by implementing monitoring, alerting, and fallback mechanisms, improving production incidents reporting and on-call efficiency by almost 40%.
• Worked with product, frontend, infra, and QA teams to ship platform features including live chat, real-time event networking, and attendee engagement tools platform services.

May 2021 – Jan 2022 · 8 mos

• As a Software Development Engineer I, I focused on building and optimizing user-facing features, improving performance, and contributing to platform stability. Key contributions include:
• Built backend services for real-time presence indicators and live viewer counts, enabling accurate session attendance tracking for hosts during live events.
• Developed backend logic for live reactions and Q&A features, handling high-frequency concurrent user interactions with low-latency response under event-scale load.
• Developed and maintained REST API endpoints for session management and attendee interaction features, ensuring reliability under high-concurrency event workloads.
• Improved landing page performance by 50% by offloading assets to CDN, preloading critical resources, and optimising imports with tree shaking.
• Wrote unit and integration tests for backend services powering critical event features, reducing regression risk during live sessions.
• Contributed to CI/CD pipeline improvements and collaborated with designers and product managers to ship user-facing features across live sessions.

Dec 2020 – Aug 2021 · 8 mos · Bengaluru, Karnataka, India

Jan 2020 – May 2021 · 1 yr 4 mos · Gurugram, Haryana, India

• As a Software engineer at 91wheels.com, I have made significant contributions to the company. Here are some of the key contribution. I joined 91wheels as first engineer and helped to develop the product from 0 to 1.
• 1. Developed the front-end of 91wheels.com from scratch, creating a user-friendly and visually appealing interface that has received positive feedback from users.
• 2. Built the REST API backend architecture of 91wheels.com from scratch using NestJS, MYSQL, Redis, and AWS, enabling the efficient processing of user requests and data retrieval.
• 3. Worked on the development of AMP pages from scratch, resulting in a significant increase in organic search traffic and improved user engagement.
• 4. Improved the web app performance, taking the Google page speed score from 9 to 70, by refactoring code, using best practices, and removing unused JavaScript.
• 5. Built the backend for a CMS to add vehicles data and lead generation using ReactJS, NestJS, MySQL, Elastic Search, Redis, and PHP, enabling the team to easily manage and update vehicle information and generate leads efficiently.
• 6. Collaborated with the design and product management teams to identify user needs, prioritize features, and deliver new functionality in a timely and efficient manner.
• 7. Mentored junior team members, conducted code reviews, training workshops and implemented best practices to ensure high-quality code.

Nov 2018 – Jul 2019 · 8 mos · Gurugram

• Built a 1:1 mentorship marketplace connecting students with professionals; scaled to 30K+ MAU.
• Solved cold-start by curating career guides & resources, driving initial demand.
• Shut down after a year due to scaling challenges.

Mar 2018 – Jun 2018 · 3 mos · Jaipur, Rajasthan, India · On-site

• Explored Different Deep Learning and Computer Vision techniques
• Worked on Implementing Action Recognition using 3D Convolution Neural Network Technique on UCF-101 dataset.
• Achieved a considerable accuracy of 76.4 %.
• https://github.com/dipakkr/3d-cnn-action-recognition

Aug 2017 – Jun 2019 · 1 yr 10 mos · New Delhi Area, India

• Selected as Microsoft Student Partner (MSP) twice from university.
• Led the college chapter of the Microsoft Student Community, building and fostering the campus tech ecosystem through events, mentorship, and peer learning initiatives.
• Organised 10+ technical workshops on and off campus, covering topics across Microsoft technologies, AI, and software development.
• Represented the community at the Microsoft Student Partner Summit at BITS Pilani, collaborating with top student technologists across India.
• Won Best Blog Award at the Microsoft AI Challenge and led a successful SMS Organizer Campaign during MSP tenure.

Jun 2017 – Jul 2017 · 1 mo · New Delhi Area, India

• Worked on building android app that can generate wallpaper and gif images. Used external API integration for this purpose.
• Awarded with Best Intern of Batch 2017.