Akshat Sharma

Resourceful and impact-driven Data Scientist with a strong foundation in AI/ML, statistical modeling, and end-to-end product development. Adept at translating complex business challenges into scalable data-driven solutions, with hands-on experience across experimental design, predictive modeling and deep learning. Proven track record of building production-ready systems—from real-time behavioral profilers to engagement and financial analytics dashboards. Combining a hacker’s creativity with a researcher's precision, always optimizing for maximum insight, minimal noise, and meaningful impact.

NeuroTrace

• Built a real-time Linux process profiler using eBPF and BPF ring buffers to trace syscalls, file I/O, and network activity at the kernel level, enabling fine-grained system telemetry.
• Trained and integrated a lightweight LSTM-based neural network via ONNX Runtime to classify process behavior (e.g., benign, malicious, etc) based on syscall patterns and memory usage.
• Deployed a live dashboard using Prometheus and Grafana with alerts for system observability.

Data Glove

• Engineered a glove with flex sensors, accelerometers, gyroscopes, integrated with Arduino R2 for real-time data processing.
• Made a novel classification model utilizing Vision Transformers and Bi-Directional LSTMs, achieving 95% gesture recognition accuracy.

CTRL+INTEL

• Designed an AI research agent integrating Perplexity Sonar and Gemini Flash 2.5 to automate deep data collection, processing up to 1M tokens of real-time web intelligence for any topic. Used PostgreSQL enabling smooth data ingestion, semantic restructuring, and persistent storage for instant client access.
• Implemented a smart retry mechanism and rate-limiting guardrails within the API workflows to prevent overloading the Perplexity endpoints and ensure fault tolerance.
• Reduced manual analysis time by 90% and transformed unstructured web data into actionable, client-ready insights.

Unigram

• Built a full-stack social media web app with real-time one-to-one and group chat using WebSockets.
• Integrated services like live streaming, pub/sum model for notifications, load balancing and reverse proxies.
• Reduced load times by 95% using Redis caching.

• AWS Academy Cloud Foundations, October 2023
• Data Analytics Professional Certificate, Google, June 2024

• Rated 5 stars in C++ on Hacker Rank
• Won the Best Delegate in VITB MUN 2024
• Achieved over 100 installations and a rating of 4.5 on my VS Code extension, CodeWhisper

• GitHub: github.com/akshats911
• LinkedIn: https://www.linkedin.com/in/akshat-sharma-4842a822a