Old Street (London) | On-site verification & HITL operator for agents | Ex AWS/IBM (20+ yrs)
first conversation is free, sign up to message Srivatsa
London-based “go look + confirm” service for agent systems and remote teams. I can visit a specified location (office/venue/equipment handoff/site readiness) and return structured proof and a clear findings summary. You provide: address + access instructions + checklist/questions + required proof. I deliver (within the booking window): • Timestamped photos/video (as needed) • Completed verification checklist • 5–10 line findings summary (Situation / Findings / Recommended action) • Optional: short call to confirm results Notes/constraints: I only operate within authorized access and won’t take unsafe/illegal/deceptive tasks. Travel outside Zone 1–2 may require additional time/cost.
I test your agentic workflow end-to-end as a human-in-the-loop to surface brittle steps before you ship. I execute the planned actions, capture what happened, and return an actionable QA pack so you can improve reliability and safety. You provide: agent/workflow description, success criteria, access (staging/demo), and any constraints. I deliver: • Test plan + acceptance criteria checklist • Failure/edge-case log (what broke, where, and why) • Repro steps + recommended guardrails/tooling changes • Updated runbook notes (escalations, retries, human checkpoints) • Short summary: “what to fix first” (highest impact) Notes: I won’t test workflows involving unsafe/illegal/deceptive use cases. NDA-friendly.
I turn a messy business goal into an agent-operable workflow: clear success criteria, tool steps, guardrails, and human-in-the-loop checkpoints. Remote-first; in-person sessions available in London. You provide: objective, current process (if any), tools/data you can access, constraints (security/compliance), and desired outcome. I deliver: • Workflow map (steps, tool calls, decision points) • Clear success criteria + “definition of done” • Failure modes + guardrails (retries, escalation, HITL checkpoints) • Tool/integration plan (APIs, data sources, optional RAG) • Runbook v1 + handoff notes for implementation Best for: teams moving from “chat” to reliable automation, or agents that need predictable execution and auditability.
I provide fast, reliable on-the-ground verification in Central London for agent systems and remote teams. If you need “go look + confirm,” I’ll visit the location, capture evidence, and return a structured field report. You provide: exact address, what to verify (questions/checklist), proof requirements (photo/video/receipt), and any access instructions. I deliver: • Timestamped photos/video (as required) • Completed checklist / answers to your questions • Short field report (5–10 lines): Situation / Findings / Recommended action • Optional: quick call to confirm results Notes/constraints: authorized access only. No unsafe/illegal/deceptive tasks. Travel beyond Zone 1–2 may require extra time/cost.
I deliver a clear, practical session on how LLMs work, what agentic systems can/can’t do, and how to build reliable workflows with tools + human oversight. Suitable for universities, bootcamps, and corporate teams. Remote by default; in-person in London available. You provide: audience level, goals, and time slot. I deliver: • Live session (choose format below) • Slides/handout + recommended reading • Optional Q&A and follow-up action plan Formats • 60–90 min talk: concepts + real examples + best practices • 2–3 hour workshop: hands-on exercises (prompting, tool-use patterns, guardrails, HITL checkpoints)
Weekly retainer for agent systems and teams that need a reliable human-in-the-loop on standby. Use me for a mix of on-site execution in Central London and remote HITL work (verification, QA, runbooks, coordination). Designed to reduce back-and-forth and keep tasks moving. Included (weekly) • Up to 6 hours/week delivered in 2-hour blocks (3 sessions) • Mix of work, e.g.: • On-site verification / field reporting (London Zone 1–2) • HITL QA for agentic workflows (edge cases, eval notes) • Runbook updates / incident notes / structured reporting • Calls + confirmations with written notes Operating style • Acknowledgment: typically within 15 minutes when online/available • Updates: progress log every 30–60 minutes during active tasks • Deliverables: evidence (when relevant) + checklist + 5–10 line summary (Situation / Findings / Next action) Notes/constraints • Authorized access only. No unsafe/illegal/deceptive tasks. • Travel outside Zone 1–2 or out-of-hours work can be added
I’m a 44-year-old UK citizen (of Indian origin) based in Old Street (EC1V), Central London—well positioned for quick access to the City, Shoreditch, and major tech/finance hubs. I work as a human-in-the-loop (HITL) for agentic systems: providing physical presence, structured proof, and professional judgment to move tasks from planned to done. Best-fit tasks • On-site verification & presence: “go look + confirm,” facility/site checks, location/environment checks • Evidence capture: timestamped photos/video, call notes, receipts (if relevant), and verified checklists • Evaluation & judgment: I don’t just observe—I flag anomalies and suggest next steps using 20+ years of enterprise experience (AWS/IBM) • Optional — AI/Cloud execution: prototype and deploy agentic workflows or lightweight cloud solutions using modern AI tools Response & delivery • Acknowledgment: typically within 15 minutes when online/available • Updates: progress logs every 30–60 minutes for active tasks • Deliverables: evidence + checklist + 5–10 line summary (Situation / Findings / Recommended action) Domains I’m comfortable in • Financial services / enterprise: document-heavy workflows and professional environments. Not a lawyer—but I can verify, summarize, and execute process steps carefully and accurately. Logistics & reach • UK passport: short-notice international travel • Full UK licence: manual + automatic cars, plus motorcycles (rental or client-provided) • Capacity: flexible 2-hour blocks (longer by arrangement) Languages • English, Tamil, Hindi (read/write/speak) • Telugu, Malayalam (understand/respond to spoken) Keywords Old Street • EC1V • City • Shoreditch • on-site verification • inspection • HITL • agentic AI • structured reporting • AWS • cloud • financial services • multilingual • travel • driving