Data Scientist & AI Engineer

ashuguptahere@gmail.com • LinkedIn • Leetcode • GfG • Codolio • Kaggle • GitHub • Portfolio

💼 EXPERIENCE

Data Scientist & AI Engineer | (Dec 2023 - Present)
Quantum Leap | Setúbal, Portugal

Leading and delivering computer vision solutions in Human Activity Recognition and Behavior Analysis to enhance retail loss prevention and identify suspicious customer behavior
Developed and deployed an end-to-end system using Ultralytics and custom computer vision pipelines
Orchestrated an in-house dataset with semi-automated labeling for real-world behavior detection
Deployed optimized models on NVIDIA Jetson AGX using DeepStream and TensorRT, reducing inference time by 40% and enabling low-latency, high-throughput edge inference
Achieved 10-20x (1000-2000%) inference speedup by migrating Python-based code to C++ DeepStream pipelines optimized for NVIDIA hardware
Automated large-scale deployment with Ansible and GitHub Actions with scalable architecture; currently running on 10+ devices with architecture scalable to 1000+ devices
Contributed to AIOps/MLOps best practices by integrating CI/CD, remote monitoring, and model versioning for reliable, scalable deployment and lifecycle management
Led the development of a fully localized Qwen-2.5/3 LLM model tailored for a law firm client, mentoring 2 junior engineers and owning delivery of production-ready features.
Built a sophisticated chat interface allowing the client to interact with their data, facilitating quick identification of relevant documents and specific information within vast repositories of PDFs and text files.
Spearheaded & successfully led initiatives in RAG & LoRA-based fine-tuning, enhancing domain-specific model efficiency
Engineered backend solutions to streamline and analyze incoming streams, enabling real-time irregular behaviors detection
Led the development of a fully localized Llama-3 70B LLM model tailored for a law firm client
Mentored junior AI Engineers and Data Scientists, enforcing strong ML engineering practices and high-quality production code.
Contributed to AI team growth by participating in technical hiring, interview evaluations, and onboarding of new engineers.

Software Engineer 1 | (Dec 2022 - Dec 2023)
MAQ Software | Noida, India

Spearheaded initiatives in Retrieval-Augmented Generation (RAG)
Applied state-of-the-art LLMs like GPT-3.5, GPT-4, and Llama 2 for RAG applications
Optimized a legacy system using Python, NumPy & Pandas, cutting runtime from 7 minutes to 10 seconds (~42x faster)
Deployed production-grade solutions on Azure: Function App, Web App, Cognitive Search (Index/Indexer), Blob Storage, SQL Databases, Speech Service
Implemented a high-impact recommendation system for ~1M users

System Engineer | (Aug 2021 - Dec 2022)
Tata Consultancy Services (TCS) | Noida, India

Led and executed numerous Data Analysis and Data Science Projects
Proficient in backend development in Python for robust, scalable applications

🧑‍💻 INTERNSHIPS

Data Science Intern | (Jul 2021 - Aug 2022)
VerSe Innovation | Bengaluru, India

Engaged in challenging Computer Vision endeavors, including tasks encompassing Image and Video Classification

AI Trainer | (Jan 2020 - Feb 2020)
CETPA InfoTech Pvt. Ltd. | Noida, India

Educated and taught college students in their 3rd and 4th years on a wide array of AI topics

Machine Learning & Deep Learning Intern | (Jun 2019 - Jul 2019)
CETPA InfoTech Pvt. Ltd. | Noida, India

Internship Involvement in Twitter Sentiment Analysis Project

🛠️ SKILLS

Technical: AI, ML, DL, GenAI, Data Science, Data Analysis, Computer Vision, Anomaly Detection, Image Encryption, Python, Git/GitHub, MySQL, T-SQL, Video Analytics, DeepStream, GStreamer, MQTT, LLM, NLP, RAG, LoRA/QLoRA fine-tuning, Vector Databases, EDA, C/C++, DSA, Linux, PowerBI, AIOps, MLOps, Docker, Prometheus, Grafana, Tailscale, Redis, AWS, Azure
Libraries: PyTorch, TensorFlow, Keras, OpenCV, scikit-learn, NumPy, Pandas, Seaborn, Plotly, NLTK, OpenAI, Streamlit, Gradio, Flask, FastAPI, LangChain, LangGraph, LlamaIndex, uv, ruff
Exposure: C#, Java, HTML, CSS, JavaScript (ECMAScript), Mojo, Rust, Zig
Languages: English, Hindi, Portuguese (beginner)

📂 PROJECTS

Real-Time Sign Language Detection : Achieved a validation accuracy of 99.348% and a minimal loss of 0.03621 on ResNet102. Dataset: ASL Alphabet
Human Activity Recognition : Conducted Video Classification on two distinct massive datasets, UCF11 and UCF101. Demonstrated robust validation accuracy ranging from 90% to 95% on ResNet102. Datasets: UCF11, UCF101
AI Cover Letter Generator : Simple and intuitive web interface made using Gradio with real-time Cover Letter Generation. Usage of Llama 3.2 model from Ollama and usage of uv package manager for package dependency.

🏆 Achievements

Gate 2021 cleared with Marks: 25.21 & Gate Score: 339
Practiced over 190+ DSA questions on various platforms

🎓 Education

Bachelor of Technology, CSE from Shaheed Bhagat Singh State Technical Campus (SBSSTC) — 2016 - 2020

Aashish Gupta