Data Science & GenAI Tech Lead — AI Agents, Structured Insights & Detection
Location: London, hybrid
About Anecdote
Anecdote we’re on a mission to make customer experience delightful for everyone involved . Think a real‑time copilot that listens to live calls and chats, reads our customers’ knowledge bases, and drafts high‑quality replies for human agents - while also turning messy, multi‑channel feedback into trustworthy structured insights, anomaly detection, and novelty discovery.
As the Tech Lead, you’ll own the technical vision and turn requirements into a live, reliable product used by brands like Grubhub, Booking.com, Dropbox, Uber, Careem, and Fubo. You’ll collaborate directly with engineers, other tech leads, directors, and the CTO to evolve ambitious prototypes into a rock‑solid, scalable platform.
What you’ll actually do
50% Build — design & ship
- Agentic AI for CX: Real‑time assistants that listen to calls/chats, retrieve from customer KBs, and draft responses with human‑in‑the‑loop controls.
- Structured extraction: Schema‑driven pipelines over unstructured text (and other modalities) using retrieval, tool‑use, and robust LLM prompting.
- Hybrid anomaly detection: Blend classical time‑series methods (e.g., decomposition, change‑point, forecasting) with LLM‑aware, contextful detectors for seasonality, spikes, step‑changes, and drift.
- Novelty discovery: Embedding‑based clustering and drift, topic surfacing, LLM summarization of emerging themes with deduplication and evidence links.
- Alerting & scoring: Severity/impact ranking, de‑noising, suppression/cool‑downs, routing, and feedback loops.
25% Architect & scale
- Own reliability, latency, and cost. Design online/offline eval harnesses, canaries, and SLAs; operate GPUs/accelerators where needed.
- Stand up and harden RAG pipelines (indexing, retrieval policies, grounding, guardrails) and agent frameworks.
- Take basic infra ownership on GCP (or AWS/Azure): networking, autoscaling, CI/CD, IaC, observability, and cost tuning.
- Participate in on‑call for your area and drive root‑cause analysis with crisp follow‑ups.
15% Collaborate
- Pair with back‑end & front‑end to wire extractors/detectors and agents into ticketing, voice, and analytics stacks (APIs, webhooks, real‑time streams).
- Partner with PMs/CX to evolve taxonomies, schemas, and guardrails; translate business problems into shipped ML features.
10% Align & showcase
- Gather requirements from CX and product leads, demo new capabilities to execs & customers, and document impact with precision/recall, alert quality, latency, and cost metrics.
What makes you a great fit
- Startup hacker mindset: You self‑start from zero, respect no silos, and carry work from prototype to production. 🛠️
- AI‑native dev tools are your daily drivers: Cursor, v0, Claude Code (or similar).
- 7–10 years building production ML/back‑end systems; 2+ years leading while coding.
- Expert Python; strong back‑end chops (e.g., FastAPI, gRPC, Postgres, pub/sub/streams).
- Agents & RAG: Fluency with at least one agent framework (ADK preferred). Proven track record shipping AI agents and building RAG pipelines.
- LLM + DS depth: Prompting/tooling, retrieval design, LLM evals; hands‑on with time‑series analysis (forecasting, change‑point, drift).
- Cloud & ops: Basic infra ownership on GCP (or AWS/Azure): networking, autoscaling, CI/CD, IaC, observability, and cost control.
- Communication: You explain results clearly, align stakeholders, and write crisp docs.
Bonus points
- DevOps wizardry; GPU/accelerator experience.
- Multimodal pipelines (text + voice + screenshots).
- Prior experience in contact center/CX analytics or novelty/anomaly systems.
- Founder or founding engineer experience
