Dheeraj Anikar
AI Implementations / Forward-Deployed AI Engineer
I deploy production AI agents (chat, voice, internal assist, QA) into real customer environments and make sure they behave. I work at the intersection of conversation design, integrations, orchestration, and post-deploy monitoring to keep LLM systems safe, auditable, and aligned with business + compliance constraints.
Experience
Eltropy Inc.
AI Implementations Engineer (Forward-Deployed)
June 2024 – Present
- Own end-to-end deployment and implementation of AI chat + voice agents for 20+ credit unions in regulated financial environments - requirements capture, conversation design, integrations, launch readiness, and customer training.
- Shipped and now support 30+ production agents (member-facing, voice IVR, internal assist, and QA/review agents), focused on measurable containment, call deflection, and first-contact resolution.
- Built and evolved an internal Agentic QA framework: orchestrated multi-agent evaluators (persona simulator, test orchestrator, factuality/verifier agent) to generate synthetic conversations, probe failure modes, and surface safety gaps before release.
- Implemented retrieval-grounded generation and verifier loops so agents stay on-policy, respect disclosure / compliance rules, and don’t hallucinate product details or rates.
- Built a sitemap-driven Q&A generator and LLM/RAG evaluation harness in Python using Gemini and Firecrawl to auto-generate and score ground-truth question/answer pairs. Deployed behind AWS services and Kubernetes for scalable benchmarking.
- Designing agentic web automation workflows using Browserbase + OpenAI: autonomous task agents that can navigate UIs, extract data, and complete multi-step flows for customers (not just answer questions).
- Built a capacity / bandwidth observability dashboard in Amazon QuickSight by ingesting GuideCX task data via API and automating ETL with AWS Lambda + EventBridge; gave leadership live visibility into delivery load and risk.
- Authored Safe AI playbooks and ran prompt-injection / jailbreak testing. Directly trained both internal teams and customer stakeholders on AI governance, escalation paths, and human-in-the-loop review.
SwingVision
Data Analyst Intern (AI & Product Insights)
May 2023 – August 2023
- Built a supervised learning model to generate match-strategy recommendations for players; improved suggestion accuracy by ~15%.
- Automated analytics ingestion using Python and REST APIs to pull behavioral and performance data from Amplitude Analytics and unify it for downstream analysis.
- Prototyped early LLM / LangChain tooling to generate narrative feedback summaries for players, moving toward automated “match debrief” coaching.
Bosch Global Software Technologies
Engineer, Safety & Risk Analysis
Aug 2020 – July 2022
- Built Python automation to accelerate functional safety analysis for automotive systems — cut recurring manual review effort enough to save roughly ~$10K/year in engineering time.
- Ran Monte Carlo simulations to stress-test component reliability across operating conditions, feeding results into design and compliance docs.
- Delivered dashboards (Matplotlib, Seaborn) to visualize risk posture and trendlines for leadership and auditors.
Skills
Applied AI / Agent Systems
LLM Agents (chat / voice / IVR)
Multi-agent orchestration
Agentic QA & verifier loops
Retrieval-grounded generation
Post-deploy monitoring & eval
ML / GenAI
LLMs / RAG
Prompt & policy design
Safety / alignment surfaces
Fine-tuning (LoRA / Q-LoRA)
Synthetic Q&A generation
Languages / Scripting
Python
SQL
Node.js (beginner)
HTML
Infra / Tooling
AWS / Lambda / EventBridge
Kubernetes
Browser automation (Browserbase)
QuickSight dashboards
Delivery / Customer Work
GuideCX & JIRA
Customer go-live & UAT
Requirements scoping
Regulated onboarding & training
Executive reporting / capacity planning