My Resume

Harshavardhan M
B.Tech in Computer Science, specialized in Artificial Intelligence
Masters in Artificial Intelligence

Education

Experience

BeSuperhuman.ai (June 2023 - August 2026, London, UK. Remote)

• Lead Machine Learning Engineer
• Developed the Minimum Viable Product (MVP), overseeing backend infrastructure and AI models utilizing cutting-edge LLMs (Large Language Models), Langchain. Was able to cut costs more than tenfold, by integrating vector databases for prompt retrieval.
• Created an autonomous agent with Langchain and LLM technologies, featuring a custom BabyAGI version that autonomously navigates browsers and emulates human inputs, resulting in improved task efficiency and reduced manual intervention.
• Engineered a cutting-edge desktop application from inception to completion, utilizing technologies such as Electron Python, Flask, and implementing a browser control module using Seleniumbase.

Agrisoft Diary and Agro Producer Company Ltd. (Jan 2022 - April 2022, Kollam, India)

• Full Stack Developer Intern
• Designed and implemented a novel ERP software pipeline.
• Worked on the development of two Flutter applications with Firebase as the backend.
• Developed a REST API using Express and created a custom user interface to access the backend.
• Designed and integrated a payment solution tailored to client needs.
• Led weekly client meetings to facilitate open communication and ensure seamless project development.

AmritaCREATE Labs (June 2022 - April 2023, Amritapuri, India)

• Deep Learning Research Intern
• An edu-tech research initiative that are funded principally by research grants.
• Working on Sign Language Accessibility for e-Governance Services at Amrita CREATE. We use AI/ML for sign language recognition of questions, that is asked to the chatbot in FAQs across 25 UMANG services.
• Developed a Deep Learning pipeline and successfully implemented a custom BERT model tailored to address this specific problem.
• Created a pioneering solution by integrating heatmaps into the pretraining process of a transformer-based model.

Projects

EazyPredict (Python Module, Python, Skikit-Learn) (Jan 2022) • Python module that provides an initial set of ML models given a classification/regression input. • Designed and implemented ensemble learning functionality using a voting classifier to automatically combine top-performing models for improved accuracy. • Integrated automated performance metrics for classification and regression tasks, simplifying model evaluation with accuracy, f1 score, ROC AUC score, RMSE, and R-squared outputs. • Optimized the module to reduce memory usage, making it suitable for resource-constrained platforms like Kaggle.

Not Blue Chrome Extension (Deep Learning/NLP) (May 2021)

• Developed a chrome extension that monitors the search activity of a user to identify potential signs of depression.
• Utilized a deep learning LSTM model to calculate a sentiment score for each search phrase.
• Implemented a threshold mechanism to detect concerning levels of sentiment, triggering an email notification to the user’s friends and family.
• Tools & technologies used: Tensorflow, Flask, Javascript

Packet Sniffer (Computer Networking) (October 2021)

• Developed a simple packet sniffing tool using Python.
• Implemented the tool to capture data on a raw socket and provided an interactive dashboard to display network usage and other relevant information.
• Worked on the Flask back-end and parts of the front end.
• Tools & technologies used: Python, Flask, HTML, Bootstrap, JavaScript, and jQuery.

Space Invaders using Deep Q-Learning (Reinforcement Learning) (May 2022)

• A bot that plays the classic atari game, Space Invaders.
• Space Invaders is a classic japanese shooting video game that was released for Atari 6000.
• This bot was trained using a convolutional neural network as a feature extractor. It was then trained using the dueling neural network strategy.
• Tools & technologies used: Python, Keras, Arcade Learning Environments

Technical Skills And Spoken Languages

• Programming: Python, Java, MATLAB, R, HTML/CSS, Javascript, Dart, Bash, Kotlin
• Tools & OS: Visual Studio, Jupyter Notebook, Google Collab, Git, Flutter, Android Studio, Gazebo
• Libraries/Frameworks: Pandas, Numpy, PyTorch, Tensorflow, nodeJS
• Languages: English, Hindi, Malayalam

Awards and Extra Curricular Activities

AMFOSS (2019)

• Open source coding club
• Contributed to open source repositories in the fields of Android app development and web development.
• Actively participated in diverse club activities, including attending tech talks and hackathons.

Ayudh Amritapuri (2019-2022)

• International Non-Governmental Organization
• Volunteer for Amala Bharatam Campaign, AYUDH India, participated in clean-up drives and organized awareness drives in 7 venues with a team of 200+ in August 2019.
• Conducted multiple webinars on “Open source software and why you should get started” for underprivileged high school graduates, helping them embark on their software journey.

AI at Amrita (2020-2022)

• Coding club for AI/ML developers
• Worked on multiple machine learning pipelines and actively participated in numerous competitions and hackathons.
• Conducted informative seminars on machine learning algorithms to support the learning and development of newer club members.

Cubers at Amrita (2020-2023)

• Speed cubing club for Rubik’s cube enthusiasts.
• Participated in and helped organize multiple international speed cubing events recognized by the World Cube Association.
• Developed strong collaboration skills within the team, facilitating the on-boarding of new club members.