Headline

Forward Deployed Engineer @ Consuma.ai | Architecting Agentic AI, RAG & LLM Systems for Production

About

I am a Forward Deployed Engineer at Consuma.ai (Founding Team), where I bridge the gap between SOTA AI research and real-world business impact. My role sits at the intersection of complex problem-solving, client collaboration, and hardcore software engineering. Currently, I architect scalable Agentic AI workflows and real-time backend infrastructure on GCP/AWS. I specialize in moving beyond simple LLM wrappers to build autonomous intelligent systems. My day-to-day involves designing multi-agent orchestration frameworks and building hybrid memory architectures (RAG + Vector/Knowledge Graphs) that power reliable, multi-step reasoning workflows. Because I operate on the front lines of deployment, I am deeply focused on the engineering realities of AI: optimizing inference latency, managing token costs, implementing Model Context Protocol (MCP) for secure tool calling, and ensuring sub-second performance via serverless architecture. I thrive in high-velocity, high-ownership environments where I can own the full lifecycle—from the initial client discovery to the final Docker container deployed in the cloud. 🛠️ Core Areas of Expertise: • Forward Deployment: End-to-End Solution Architecture, Rapid Prototyping, Client/Stakeholder Collaboration. • Agentic & Generative AI: Multi-Agent Systems, LangGraph, RAG, Knowledge Graphs, Tool Calling (MCP). • Backend & MLOps: Python (FastAPI), Docker, Microservices, Inference Optimization (Quantization/Pruning). • Cloud Infrastructure: GCP, AWS, Serverless Architecture, Vector Databases (Pinecone, Chroma). • Deep Learning & Vision: PyTorch, Transformers, CNNs, Edge AI. Always happy to connect with fellow builders, founders, and tech enthusiasts! email - someetsahoo654@gmail.com

Location

Greater Bengaluru Area

Work experience

Forward Deployed Engineer at Consuma, Dec 2024 - Present · 1 yr 5 mos, Bengaluru, Karnataka, India, As a Forward Deployed Engineer on the founding team, I own the end-to-end lifecycle of deploying advanced GenAI solutions. I work directly at the intersection of complex client requirements and hardcore AI engineering, ensuring that our models don't just work in notebooks, but drive actual business value in production. • Agentic AI Architecture: Designing and deploying Multi-Agent Orchestration frameworks. I build stateful agents capable of multi-step reasoning, planning, and autonomous tool execution via Model Context Protocol (MCP). • Production RAG Systems: Architecting highly accurate Retrieval-Augmented Generation (RAG) pipelines backed by Vector Databases (Pinecone, Chroma) and Knowledge Graphs to ground LLM decisions in enterprise data. • Backend & Deployment: Building scalable microservices using FastAPI and containerizing workloads with Docker. I deploy and manage these serverless architectures on GCP and AWS, ensuring sub-second latency for real-time operations. • AI Observability: Implementing observability for end-to-end tracing, allowing me to debug agent reasoning loops, optimize token usage, and catch hallucinations before they reach the client.

Research intern at Indian Institute of Technology (IIT) Goa, Jul 2023 - Oct 2023 · 4 mos, Goa, India, Conducted first-principles research in deep learning, focusing on the intersection of neural networks and complex mathematical modeling. This role built my foundation in model optimization, gradient flow, and custom neural network architectures. • Physics-Informed Neural Networks (PINNs): Developed custom PINNs from scratch using PyTorch to solve complex non-linear Partial Differential Equations (PDEs) (e.g., Navier–Stokes, Burgers’, Heat equations). • Model Optimization: Benchmarked model architectures and applied Stochastic Gradient Descent and custom loss function optimization techniques, significantly reducing computation time while maintaining high-fidelity accuracy. • First-Principles AI: Gained deep, mathematical understanding of neural network behavior, stepping outside of standard "black box" libraries to manipulate gradients and convergence manually.

Data analysis intern at Indian Oil Corp Limited, Dec 2022 - Feb 2023 · 3 mos, Paradeep, Project:- " PREDICTING THE PRODUCT PROPERTIES OF FCC MAIN FRACTIONATOR USING MACHINE LEARNING". - Data Collection was the main part - Data Cleaning and EDA for refining the Dataset collected. - Using ML model for predicting the YIELD, ASTM 95% value, Benzene and Sulphur content in final product of FCC MF( Dry Gas, LPG, LCN, MCN, HCN, LCO, CLO)

Research Intern at CSIR India, Jun 2022 - Aug 2022 · 3 mos, Bhubaneswar, Odisha, India, - Performed through research on Material synthesis - I got to know a lot about supercapacitor materials and their application in various field and also how can they replace traditional energy storage devices - Learned a lot about synthesis of nanoparticles mainly through hydrothermal and Chemical vapor deposition.

Research Intern at National Aluminium Company Limited - NALCO, Mar 2022 - May 2022 · 3 mos, Bhubaneswar, Odisha, India, - Learned about various Alumina Extraction processes and also successfully completed projects on “Extraction of Alumina from PLK” and “Defluoridation of Water using Activated Alumina”. - Learned to use scientific instruments like XRF, XRD.

Education

B.tech - M.tech dual degree, Chemical Engineering, Institute Of Chemical Technology

DAV public school, cuttack, Odisha

Public links

github: https://github.com/Someet-Git