AI Architect

About ONDC

The Open Network for Digital Commerce (ONDC) is an interoperable, protocol-based commerce network for India. By unbundling the storefront from the buyer experience, ONDC is opening digital commerce to millions of small sellers, logistics providers, and specialized service platforms across retail, mobility, financial services, agriculture, and health. We are now investing deeply in applied AI to make the network smarter for buyers, sellers, and Network Participants — from intent understanding and catalog enrichment to dispute resolution, fraud signals, and agentic shopping experiences.

The Role

We are looking for a hands-on AI Architect to lead the design and productionization of agentic AI

systems on ONDC. You will own the architecture for LLM- and SLM-powered agents that operate across the network, drive the training, distillation, and domain adaptation of small language models tuned for Indian commerce, and stand up the MLOps backbone that takes these systems

from prototype to production at network scale.

This is a senior individual contributor role with significant scope: you will set architectural direction

across multiple workstreams, mentor a growing engineering team, and partner with policy, product,

and Network Participants to ship AI capabilities responsibly and in the open.

Key Responsibilities

Agentic Applications

● Design and build multi-agent systems for buyer- and seller-side use cases — intent capture,

search and discovery, negotiation, order orchestration, issue and dispute resolution — that

operate over ONDC's protocol APIs.

● Define patterns for tool use, planning, memory, retrieval, and guardrails; establish standards for

agent observability, evaluation, and safe rollout.

● Architect retrieval-augmented systems over heterogeneous catalog, policy, and transaction

data, including hybrid lexical + semantic retrieval and multilingual indexing for Indian languages.

Small Language Models — Training, Distillation & Adaptation

● Lead end-to-end training pipelines for SLMs : data curation, tokenizer choices, supervised fine-tuning, DPO/RLHF, and evaluation.

● Drive knowledge distillation and quantization strategies (LoRA/QLoRA, GPTQ/AWQ, speculative

decoding) to ship cost-efficient models suitable for edge and low-latency network use cases.

● Build domain adaptation playbooks for Indic languages, code-mixed inputs, and commerce-specific tasks; partner with linguistics and data teams on high-quality instruction and preference datasets.

● Establish rigorous offline and online evaluation harnesses, including task-level benchmarks, red-teaming, and bias/safety testing.

MLOps & Platform

● Define the reference MLOps stack: feature and embedding stores, training orchestration, experiment tracking, model registry, CI/CD for models, and reproducible serving (vLLM/TGI, Triton, KServe).

● Stand up scalable inference for batch, real-time, and streaming workloads across GPU and CPU fleets; own latency, throughput, and unit-economics targets.

● Implement observability for prompts, traces, costs, drift, and outcome quality; close the loop between production telemetry and model improvement.

● Codify responsible AI practice — data lineage, PII handling, model cards, and review gates — aligned with ONDC's open-network and DPI principles.

Architecture Leadership

● Author RFCs, reference designs, and reusable blueprints adopted across teams and Network Participants.

● Mentor senior engineers; raise the bar on code quality, evaluation rigor, and production discipline.

● Represent ONDC in working groups, open-source efforts, and external forums on AI, DPI, and digital commerce.

What You Bring (Must-Have)

● 8–12 years of software/ML engineering experience, with at least 3 years shipping production ML or LLM systems at meaningful scale.

● Demonstrated depth in modern transformer architectures, training stacks (PyTorch,

DeepSpeed/FSDP, Megatron, Hugging Face), and distributed training.

● Strong grasp of agentic patterns (planning, tool use, RAG, evaluation) and associated

frameworks (LangGraph, LlamaIndex, DSPy, or equivalent in-house).

● Production MLOps experience: model serving, vector stores, feature stores, containers,

Kubernetes, CI/CD, and cloud (AWS/GCP/Azure).

● Excellent system design judgment; able to reason about correctness, scale, cost, and failure

modes simultaneously.

● Clear written communication; comfortable writing RFCs and influencing without authority across

cross-functional and external stakeholders.

Nice to Have

● Experience with Indic NLP, multilingual tokenization, or low-resource language modeling.

● Contributions to open-source AI/ML projects or published research.

● Familiarity with ONDC, Beckn protocol, UPI, ULI, or other India DPI rails.

● Exposure to safety/evaluation tooling (Ragas, TruLens, Lakera, Guardrails) and policy

frameworks for responsible AI.

● Background in commerce, search/recommendations, or fraud/risk ML.

Why ONDC

You will work on AI that touches a population-scale public network — not a single company's stack.

Your designs will influence how millions of small sellers and buyers experience digital commerce, and

the artifacts you build (specs, models, tooling) will largely live in the open. If you want your work to

compound across an ecosystem rather than a product, this is the seat.

How to Apply

Send a CV and a short note (300 words is plenty) on a production AI system you architected — what

you optimized for, what you traded off, and what you would do differently — to careers@ondc.org

with the subject line “AI Architect — Agentic & SLM”.

ONDC is an equal opportunity employer. We hire on the basis of merit and build teams that reflect the diversity of the network we serve.