Description

I help data-heavy companies in financial services, fintech, and regulated SaaS make agentic systems reliable enough to trust for decisions, not systems that give different answers to the same question twice.

The typical ask I take on: "this process is slow, manual, or not scaling,can AI fix it?". I own it end-to-end: from discovery and architecture through to a production system on your existing stack (Databricks, Snowflake, or Microsoft), including the evaluation and governance approval that gets it past enterprise review.

Background: 10 years in data engineering and analytics (domain focus: wealth management), focused entirely on governed, decision-safe agents using LangGraph, Databricks, Snowflake, and Python.

Recent proof: at a US broker-dealer I shipped a production analytics agent that reduced a 3-day insight to 3 minutes and passed enterprise governance — open-sourced as a Databricks Solution Accelerator: github.com/mdinu-hash/analytics_agent_databricks

Based in Romania, available for remote engagements across Europe and the US.

Industry field of expertise

Languages

English
Native or bilingual
German
Fluent

Workplace preferences

Remote only

Primarily works remotely

Pfizer
AI Solutions Engineer
PHARMACEUTICALS INDUSTRY
May 2026 - Today (3 months)
Building agentic chatbot used by clinical operations leads at a global pharma company for self-service analytics.

- Raising agent accuracy to production-grade levels, taking the chatbot from pilot to trusted daily use.
- Consolidating fragmented Genie spaces into a single governed space, reducing question-routing errors.
Databricks MLFLow Genie LangGraph Generative AI
TopTal(Remote)
AI Engineer
BANKING AND INSURANCE
January 2025 - May 2026 (1 year and 4 months)
Developed production-grade conversational analytics agent for financial services

Architecture:
- Multi-agent orchestration using LangGraph
- Stateful workflows tracking conversation history and intermediate execution steps
- Tool calling with orchestrator labeling questions into scenarios and routing to appropriate tools
- ChatDatabricks integration with structured outputs
- Observability tracking all answers and intermediate steps for debugging

Evaluations:
- Evaluation framework scoring across 5 dimensions: Clarification, No Data Acknowledgment, Accuracy, Interpretation, Transparency

Decision-Safety Features:
- Disambiguation asking clarifying questions ("Which revenue—gross, net, recurring?")
- Revealing assumptions (e.g., "Revenue: $5M – excludes inactive accounts")
- Schema-grounded analysis preventing hallucinations
- Clarification step before query generation handling analytical intent ambiguity

Results:
- Consistent answers across contexts
LangGraph Databricks MLFLow Snowflake Cortex Agents
TopTal(Remote)
Senior Business Intelligence Consultant
BANKING AND INSURANCE
December 2021 - December 2024 (3 years)
Built analytics solutions on Snowflake & Power BI enabling data-driven decisionsacrossthe organization.

Key results: increased user engagement from 15% to 40% through CRM/analytics integration, and helped meet 90% of SLA targets by uncovering process bottlenecks across 1M+support cases.
SQL Snowflake Cloud Integration Microsoft Power BI