Shailesh Pranav
Software Engineer | Generative AI & Enterprise ML Solutions
Results-driven Software Engineer specializing in Generative AI and machine learning with a proven record of deploying enterprise-grade systems that reduce inference latency by 30%, improve throughput by 25%, and maintain 99.9% uptime.
About Me
Welcome! I'm Shailesh, a Software Engineer deeply passionate about harnessing the power of Generative AI and Machine Learning to create impactful, enterprise-grade solutions.
Currently, as a Data/ML Engineer at WorldLink-US, I lead the design and deployment of real-time speech interaction systems and optimized LLM inferencing, supporting 30+ concurrent users and accelerating deployment cycles by 40%. My background includes a Master of Engineering in Robotics from the University of Maryland, providing a strong foundation in intelligent systems and robust engineering principles.
My focus is on building solutions that not only perform exceptionally well but also prioritize data privacy and security. From developing secure chatbots with role-based access control to architecting Retrieval-Augmented Generation (RAG) systems across audio, video, and text, I thrive on translating complex requirements into deployable, scalable applications.
Professional Journey
Data/ML Engineer
WorldLink-US | Nov 2023 - Present- Spearheading advancements in real-time AI interactions and performance optimization.
- Engineered real-time STT/TTS using MeloTTS & Whisper.
- Developed a multimodal chatbot (audio, video, text) for RAG.
- Optimized LLM inference with TensorRT, reducing latency.
- Containerized applications with Docker & Kubernetes.
Data Engineer (Contract) – ML Pipelines
WorldLink US / Global Agricultural Equipment Manufacturer | 2024 - Present- Implemented production-grade data engineering solutions to improve ML training reliability and model performance.
- Outlier detection & data quality: implemented outlier detection and removal logic using PySpark across large-scale datasets, improving downstream model accuracy and reliability.
- Production-grade ML pipelines: built and hardened scalable data preprocessing pipelines in Databricks, streamlining the end-to-end ML training workflow and aligning outputs to model input requirements.
- Feature engineering & model improvement: analyzed data source characteristics and refined feature inputs to support iterative model improvements and tighter feedback loops between data engineering and data science teams.
Software Engineer (Contract)
WorldLink US / Multi-brand Consumer Services Platform | 2023- Developed a highly secure document management system.
- Implemented custom RBAC with JWT authentication.
- Integrated row-level security in PostgreSQL.
- Automated workflows, saving 20+ hours/week manually.
Software Engineer (Contract)
WorldLink US / Multi-brand Consumer Services Platform | 2023- Led development of a secure, high-performance enterprise chatbot.
- Deployed secure on-premises Generative AI chatbot.
- Designed RAG architecture using Milvus vector DB.
- Reduced response latency by 25-30% via pipeline optimization.
Research Intern
NYU Tandon School of Engineering | Feb 2020 - May 2020- Contributed to assistive technology research in the Mechatronics, Control and Robotics Lab.
- Designed & fabricated a robotic rehabilitation device.
- Developed embedded C++ & Python control software.
- Implemented real-time sensor processing & motor control.
- Created modular software components (40% dev time reduction).
- Built an automated testing framework for calibration.
Featured Projects
Enterprise AI Chatbot
Built a retrieval-augmented chatbot with Mixtral/Llama 3, Neo4j and Milvus to support scalable enterprise AI workflows and reduce response time for data-driven user inquiries.
Human Avoidance Robot Module
Developed a vision-based human detection module that reliably tracks people and transforms coordinates for safer robot navigation in dynamic environments.
Self-Driving Trajectory Generation
Implemented a Vision Transformer-based planner for dynamic driving scenarios, improving trajectory prediction accuracy using the Lyft Level5 dataset.
Urban Search & Rescue Sim
Designed a two-robot simulation to map victim locations and coordinate rescue actions, demonstrating effective multi-agent search and path planning in emergency scenarios.
DFS Maze Solver Robot
Created an autonomous maze-solving robot using Depth First Search, demonstrating reliable path planning and navigation in a micro mouse simulation.
Technical Expertise
Programming & Scripting
- Python
- C/C++
- C#
- Java
AI / Machine Learning
- TensorFlow, PyTorch, Keras
- Langchain, Llama-Index
- TensorRT Optimization
- Generative AI (LLMs, RAG)
- Computer Vision, NLP
Databases & Vector Stores
- PostgreSQL, Redis
- Neo4j (Graph DB)
- Milvus, Chroma DB (Vector)
DevOps & Infrastructure
- Docker, Kubernetes
- FastAPI
- Git, GitHub Actions, Jenkins
- AWS/Azure/GCP (Basics)
Other
- Robotics (Simulation, Control)
- Embedded Systems
- Unity (VR/AR Dev)
- Agile Methodologies
Education
University of Maryland
Master of Engineering: Robotics Engineering
May 2023
College Park, MD, USA
PSG College of Technology
Bachelor of Engineering: Robotics and Automation Engineering
May 2020
Coimbatore, TN, India
Let's Connect!
Interested in collaborating or discussing opportunities in AI, ML, or Software Engineering? I'd love to hear from you.