AI Architect

As a Principal AI Architect at FourKites, you will own the architecture underneath every FourKites AI agent, including the harness they run in, the context they see, the models they call, and the evals that measure whether they work. This is a hands-on individual contributor role based in Chennai (hybrid). You will write code most weeks, lead design reviews across agent teams, and make decisions on model choice, context strategy and evaluation that every team builds on. You will work closely with agent teams in Chennai and with colleagues in the US and Europe.

Responsibilities

Define the reference architecture for FourKites agents (agent loop, tool execution, state and checkpointing, retries and recovery, guardrails, permissions, human-in-the-loop handoffs); Design for long-running, multi-step work that spans many context windows and recovers when a tool, carrier API or model call fails; Decide where each problem is best solved (frontier LLM, open-weight model, fine-tuned encoder, or plain code) and own model routing, cost-per-task and latency budgets; Integrate agents with FourKites' event-driven platform (Kafka, microservices, Postgres) and customer systems through APIs and MCP servers; Set the standard for agent context windows within an explicit token budget; Build shared components for retrieval, compaction and summarization, sub-agent isolation, prompt caching and persistent memory; Diagnose context failures and turn fixes into reusable harness patterns; Build the offline eval framework (golden datasets, code-based checks, LLM-as-judge graders, CI regression suites); Build online evaluation (production traffic scoring, trace- and task-level metrics, drift detection, shadow and A/B rollouts, user feedback signals, quality alerts); Connect eval metrics to business outcomes such as tracking accuracy, exceptions resolved without a human, and appointment fill rate; Review designs and PRs for agent teams, write architecture docs and playbooks, and mentor staff and senior engineers; Track the model and tooling landscape and give leadership build, buy or wait recommendations; Partner with Product, Solutions and Customer Success to turn field failures into platform fixes

Requirements

Architected and operated at least one LLM agent system with real users and real traffic (tool calling, multi-step workflows, failure handling, human escalation) and owned it after launch through incidents, cost and latency; Deep ML knowledge of encoders and transformers (attention, positional encodings, encoder-only vs decoder-only vs encoder-decoder, tokenization, KV cache); Trained or fine-tuned encoder models (BERT-family, sentence embeddings, bi-encoders, cross-encoder re-rankers) for classification, extraction or retrieval, and understands how contrastive training, distillation and quantization change cost and quality; Context engineering: managed token budgets, built retrieval and memory for agents, used compaction and sub-agents, and measured how quality changes as the window fills; Harness engineering: built the agent loop, tool dispatch, state and checkpoints, interruption and recovery, sandboxing and permissions, guardrails and model routing; Fluent with MCP, structured outputs and streaming; Hands-on online and offline evals: datasets, graders, LLM-as-judge calibration, CI regression gates, production scoring, trace sampling, drift monitoring, shadow deployments and feedback loops; Used evals to block a release or justify a model switch; 10+ years building software or ML systems, with 3+ years building on LLMs; Strong hands-on Python, comfortable reading and reviewing Java or TypeScript services; Production experience on AWS or GCP, containers and Kubernetes, and event-driven systems; Excellent oral and written communication skills

Nice to have

Experience in supply chain, logistics or transportation; Document AI: OCR, layout models, and extraction from bills of lading, invoices and PODs; Voice or email agents that act on behalf of a business; Agent security: prompt injection defence, tool permissioning, and tenant data isolation; Hands-on experience with several model families (Claude, GPT, Gemini, open-weight models) and serving stacks; Published work, talks or open-source contributions on agents, retrieval or evaluation

Benefits

Stock options; Medical benefits starting on first day of employment; 36 PTO days (Sick, Casual and Earned); 5 recharge days; 2 volunteer days; Home office set-up and technology reimbursement; Lifestyle and family benefits; Parental leave for all parents; Annual wellness stipend; Mental wellness support and guidance; Ongoing learning and development opportunities (professional development program, Toastmasters club, etc.)