Ram Vinjamuri

Welcome to my portfolio

About Me

I am a Forward Deployed Software Engineer at Palantir with a particularly strong background in Computer Science from the University of Cambridge, where I specialized in deep learning research spanning mathematical theory and practical applications. During my studies, I developed a mathematical framework to compare and extrapolate graph neural network performance from small data subsets to large-scale datasets, enabling optimal model selection for specific applications. My research also involved deep dives into interpretability across cutting-edge architectures including CNNs, graph neural networks, and transformers.

My dissertation focused on dependency parsing of Vedic Sanskrit, combining my technical expertise with linguistic analysis to tackle challenging computational problems in an ancient language. I am currently in the process of publishing a paper demonstrating how I surpassed existing state-of-the-art benchmarks in language parsing using the latest models.

I leverage my Cambridge education and technical expertise in my professional work at Palantir, where I contribute to impactful projects and continue to grow as a problem solver and innovator, creating solutions that address significant real-world challenges.

My Education

University of Cambridge

Computer Science Graduate - First Class Honours

Graduated with a First Class degree, ranking 22nd out of approximately 125 students (top 15% of the cohort).

Dissertation: Dependency Parsing of Vedic Sanskrit - combining computational linguistics and machine learning techniques to analyze ancient Sanskrit texts. This research is currently in the process of being published as an academic paper, demonstrating how I surpassed existing state-of-the-art benchmarks in language parsing.

Final Year Modules & Research

My final year focused on advanced topics including Natural Language Processing, Deep Neural Networks, and Type Systems. For DNNs, I conducted research into subgroup generalisation of Graph Neural Networks, exploring how these models perform across different data distributions. In NLP, I worked with statistical models to develop embeddings and implemented various parsing techniques including dependency and constituent parsing for English. This research directly informed my dissertation work on Vedic Sanskrit parsing.

Extra & Super-Curriculars

During my time at Cambridge, I played for the 2XI of the Cambridge University cricket team, representing the University at 3 varsity matches and captained the King's College cricket team. I was also vice captain of the college table tennis team.

I founded the Turing Society at my college to foster a community of coding enthusiasts and promote computational thinking. I also served on the committee for CUCATS (Cambridge University Computer Science Society), helping to organize events and initiatives for computer science students across the University.

I demonstrated exceptional technical abilities by winning first place in the Encode AI Hackathon and placing in the top 5 at the Easy A Hackathon. Prior to university, I achieved 4 A*s in Computer Science, Physics, Maths and Further Maths, and was a three-time finalist in the PA Pi competition.

I actively volunteered in events for the Hindu Society and the wider Cambridge community. A notable achievement was organizing the University's Holi celebrations, which successfully bridged the gap between students and permanent residents in Cambridge.

My Professional Experience

Forward Deployed Software Engineer @ Palantir

Currently working as a Forward Deployed Software Engineer at Palantir Technologies, where I apply my deep learning expertise and Cambridge Computer Science background to solve complex real-world problems.

Software Development Intern @ Amazon

During my 12-week internship with the Alexa team, I leveraged my Cambridge Computer Science training to build a sophisticated self-service evaluation system that optimized cross-team performance. My academic background in algorithms and data structures enabled me to deliver a solution that significantly improved team efficiency.

Research Developer @ Cambridge Chemistry Laboratory

Selected for the prestigious Undergraduate Research Opportunities Programme, I led the development of an optimized system for a cutting-edge microscope, implementing GPU-accelerated algorithms that increased image processing speed by 5x while maintaining precision. Working with non-technical researchers, I effectively translated complex computer science concepts into practical solutions and provided comprehensive documentation to ensure project sustainability.

GitHub Link

Software Engineering Intern @ Wonkknows

Applied my Cambridge Computer Science knowledge to develop a full-stack tutoring web application for educational cheatsheet creation. Demonstrated strong technical implementation skills while ensuring the project aligned precisely with management requirements.

Implemented a robust backend architecture using MongoDB and Flask, showcasing the database design principles and web development expertise gained during my Cambridge degree.

GitHub Link

Significant Projects & Academic Achievements

Elementary - Award-Winning AI Project

Leveraging the advanced AI principles studied at Cambridge, I developed a project that won the Encode AI Hackathon bounty by Virtual Protocol ($3000). This achievement demonstrates the practical application of my Cambridge Computer Science education.

I designed and implemented a Sherlock-inspired puzzle game that delivers a unique experience with each playthrough. The system dynamically generates complete scenarios including plots, characters, and environments, with over 100 tested variations.

Applying computational theory principles from my Cambridge studies, I created an adaptive difficulty system that personalizes the experience based on user interaction patterns, showcasing the practical application of my academic knowledge in machine learning and user experience design.

For more information and the submission video please refer to: GitHub Link

Cambridge Machine Learning Laboratory Research

During my time at Cambridge, I completed advanced research in the prestigious Cambridge ML Lab during the Michaelmas term. This specialized training complemented my formal Computer Science education with cutting-edge expertise in transformer architectures—the foundation of modern large language models like GPT.

Through this Cambridge research opportunity, I mastered professional ML development tools including PyTorch and Weights & Biases, implementing sophisticated transformer micro-architectures that demonstrated the theoretical concepts from my degree program.

Building on my Cambridge education, I extended my research into reinforcement learning, applying the rigorous mathematical and computational principles emphasized in the Cambridge Computer Science curriculum to solve complex AI optimization problems.

Advanced Machine Learning Specialization

To complement my Cambridge Computer Science education, I completed Andrew Ng's prestigious Machine Learning specialization. This additional training reinforced the mathematical foundations taught at Cambridge, providing a comprehensive understanding of traditional ML models from first principles.

The mathematical rigor of this specialization aligned perfectly with Cambridge's approach to Computer Science education, covering advanced techniques including neural networks and support vector machines implemented in MATLAB. This combination of Cambridge's theoretical foundation with specialized ML training has given me exceptional versatility in applying computational solutions to complex problems.

Certificate Link

Additional Technical Projects

Enviro-Chain - Blockchain Innovation

Drawing on my Cambridge Computer Science expertise in distributed systems, I developed Enviro-Chain, a sophisticated decentralized platform for environmental monitoring and optimization. This solution helps residential and commercial spaces improve environmental quality and energy efficiency by integrating real-time environmental parameters with blockchain technology for data integrity and security, while implementing machine learning algorithms for predictive analytics and optimization.

Applying the theoretical foundations from my Cambridge education, I engineered this solution using the Polka-chain framework with two primary technical innovations: implementing sharding techniques for enhanced scalability and leveraging blockchain's immutable properties for secure data storage.

This project earned recognition by placing in the top 5 at the prestigious Easy A Polka Chain Hackathon, demonstrating the practical application of my Cambridge Computer Science education. For more technical details, please visit: GitHub Link

Whatsapp Events - Intelligent Notification System

Applying the software engineering principles and system design methodologies from my Cambridge Computer Science degree, I architected a sophisticated notification system with a robust Python backend deployed as a Linux service, complemented by an intuitive Flask-based frontend. The system demonstrates advanced API integration by communicating with GreenAPI through HTTP requests for WhatsApp connectivity, while implementing natural language generation using GPT-3.5 Turbo to create contextually appropriate messages. This project showcases my ability to apply Cambridge's theoretical foundations to practical software solutions:

Optimal Reminder Times: GPT predicts the ideal times for sending reminders, eliminating the need for manual selection and ensuring maximum effectiveness.
Engaging WhatsApp Prompts: The language model generates captivating WhatsApp messages with just minimal event details.
For example, from basic information about the event location and name, GPT crafts engaging prompts like:

Reminder: Get ready for Cricket Practice at Fenners in just 30 minutes! Remember to bring your A-game and your sense of humor. As a bonus joke for motivation: Why did the cricket team go to the bank? To get their bowlers back! 😉🏏 #GameOn #CricketTime
User Input Sanitization: GPT assists in sanitizing user inputs to ensure smooth server functionality. We restrict occurrence frequency inputs to a finite set, predicting values from other cases to enhance user experience.

For more information please refer to the GitHub Link

Intellectual Pursuits & Personal Development

Classical Piano: Achieved Grade 8 certification, reflecting the discipline and precision that complements my Cambridge Computer Science education

Current Reading: "Thinking Fast and Slow" by Daniel Kahneman - exploring the cognitive science principles that inform both human and artificial intelligence decision-making processes

Research Interests: Particularly engaged with "Scaling Synthetic Data Creation with 1,000,000,000 Personas" - a paper that aligns with my dissertation work on computational linguistics and natural language processing

Academic Interests: Continuing to explore the intersection of linguistics and computer science, building on my Cambridge dissertation on dependency parsing of Vedic Sanskrit