Skip to content

Ramsi Kalia

MBA, MSc, B.E.
New Delhi, India • LinkedIn • X/Twitter • GitHub • Academia.edu • Email

AI Engineer • Agentic AI • Multi-Agent Systems • Cloud Deployment • RAG

Summary

AI Engineer (MBA, MSc) specializing in agentic AI workflows, MCP toolchains, and scalable deployment. Experienced in designing multi-agent systems that integrate LLMs, retrieval, tool use, and conversational control. Built projects including an MCP augmentation server with optional VLM integration, a dual-agent orchestration pipeline, a domain knowledge RAG system, and research on turn detection for AI-powered language learning.

AWS Certified Cloud Practitioner and AI Practitioner, with 2 years of portfolio work across deep learning, LLMOps, and MCP deployment. Strong documentation and cross-functional collaboration skills, converting research concepts into reliable, production-ready workflows.

Core Skills

Agentic AI: Designing agent reasoning pipelines, function-calling and tool use, multi-agent orchestration, retrieval-augmented generation (RAG), LlamaIndex and LangChain, prompt engineering

LLM & GenAI: LLM experimentation, model selection, embedding workflows, semantic search, tuning and evaluation for accuracy, latency, and cost

MCP & Tooling: Custom server development, tool integration, cross-platform agent interoperability, hooks and testing

Deployment: Docker, FastAPI, TorchServe, real-time/streaming inference, microservices architecture

MLOps & Cloud: DVC, MLflow, WandB, Git, GitHub, AWS (SageMaker, EC2, S3), CI/CD pipelines, certified Cloud Practitioner + AI Practitioner

Work Experience

Data Scientist - Computer Vision R&D, CarScan

2021 – 2022 | Remote - Agile, Kanban | Insurance AI | Reporting to CTO

A firm at the intersection of AI and the automotive insurance industry with ~100 employees and clients like Telesure, GT Motive, Vieva, and Allianz, operating in South Africa, Nigeria, Kenya, Ghana, India, and the Middle East.

  • Deployed a 10-class car image classifier (>96% accuracy, <300ms inference) in TensorFlowJS; reduced model size to ~3MB
  • Replaced U2-Net with DIS for salient object segmentation; cut training loss to 0.06 after 200k+ iterations and deployed via TorchServe/Flask
  • Built barcode segmentation model (Mobilenet v2/v3) compressed from 22MB → 1.2MB, achieving 96.5% IoU on live video frames
  • Created car color detection algorithm using Detectron2 + OpenCV; deployed as a microservice with Flask, Docker, and AWS
  • Scaled datasets from <800 → 50k+ images using a Selenium + AWS scraper and augmentation pipelines
  • Advocated and implemented WandB + DVC for reproducibility; introduced Cookiecutter + Sphinx for repo hygiene
  • Authored internal technical papers and experiment reports to standardize methods and share findings across R&D
  • First in company to achieve Agile certification; mentored team on sprint planning, retros, and story estimation
  • Proposed and led internal process upgrades (experiment tracking, version control, documentation standards) adopted team-wide

Technologies: PyTorch, TensorFlow, TensorFlowJS, Detectron2, OpenCV, TorchServe, Docker, Flask, AWS EC2/S3, DVC, WandB, Label Studio, SuperAnnotate

Skills: Image classification, object detection, segmentation, model compression/quantization, 3D vision, real-time inference, Agile facilitation, technical documentation

Project Manager – Engineering & Manufacturing, Mellcon Engineers

2013 – 2019 | Delhi NCR - JIT, Kaizen | Industrial Manufacturing | Reporting to MD

A Manufacturing firm with ~150 employees engaging in complete design and manufacture of industrial equipment including compressed air treatment systems and refrigeration equipment, with clients like Alstom, GE India, Indure and Bhabha Atomic Research Centre.

  • Managed 20-person cross-functional team across design, manufacturing, QA, and delivery; oversaw 10+ concurrent projects/quarter with portfolios up to $400K
  • Owned full project lifecycle: P&ID analysis, BOM creation, mechanical design, vendor coordination, QA/QC testing, and client delivery
  • Ensured compliance with ISO 9001 and BPVC standards; audited and validated production pipelines
  • Spearheaded 5S implementation and lean manufacturing across shop floor and inventory system; generated $10K+ in cost savings within 3 months

Technologies: ERP, CRM, mentoring, AutoCAD, SolidWorks

Skills: Project lifecycle management, cross-functional leadership, compliance & audits, Lean/5S, stakeholder communication, vendor coordination

Education

Master of Business Administration (MBA) | Quantic School Of Business and Technology

2023

Dissertation: Burberry APAC Market Strategy — Digitalization & ESG for Gen Y Consumers

  • Designed omnichannel strategy and segmentation plan for $20B Chinese luxury market, integrating ESG and cultural frameworks

Master of Science in Mechanical Engineering | University of Exeter

2017

Dissertation: Lean Manufacturing Techniques to Reduce Electrical Wastage in Food Processing SMEs

  • Collaborated with a UK SME to pilot Lean interventions, cutting electrical waste through Kaizen and takt-time optimization frameworks

Additional Information

Dissertation Links:

Languages: English, Korean-B2

Certifications:

  • Professional Scrum Master™ I (PSM I)
  • AWS Certified Cloud Practitioner
  • AWS Certified AI Practitioner

Community:

  • Language Cafe Discord Server Dev Team and STEM channel waiter (15k members) | Language learning community
  • 100 Days Of Cloud Discord Server Mod and Dev Team (6.5k members) | Cloud learning community

Download PDF version: Resume.pdf