Madhukar Reddy V
Founder & Principal AI and Analytics Lead · CFA Level III Candidate
Thirteen years building analytics, business intelligence and data engineering for banks, retailers and global technology firms — Microsoft, Sony, Wells Fargo, Fidelity Investments, Hudson's Bay, Sobeys and Royal Bank of Canada among them.
Today I run the analytics and AI stack for a multi-market European fintech: Azure and Databricks pipelines through to Power BI, generative AI in production, and models for risk, segmentation and fraud. Most of what I know about making AI trustworthy came from having to answer for the numbers afterwards, in front of people whose job it is to disagree with them.
ProAnalytiqs exists because the same gap kept turning up: companies who needed a senior AI and analytics team and could only hire one person at a time.
Four things, in this order
The order matters more than the list. Context before agents, agents before autonomy, and guardrails from the first commit rather than the last sprint.
Context engineering
Conformed dimensional models with declared grain, metrics defined once in dbt, and descriptions written so a machine can read them. This is the part most AI projects skip and then fail on.
Agentic systems
Agents that monitor a business autonomously — decomposing a movement into its drivers before escalating, and refusing rather than guessing when the model cannot answer.
Guardrails and evals
Golden sets, LLM-as-judge scoring, grounding checks and tracing, so a prompt change cannot silently break eleven answers that used to be right.
Decision science
Scoring, fraud, forecasting and segmentation models benchmarked against a boring baseline, monitored for drift, and documented the way a risk committee needs to read them.
Where it came from
Thirteen years across consulting, in-house and client-side delivery, in four regulated industries.
-
BI & Analytics Tech Lead — Multi-market European fintech
Owns the analytics and AI stack across eight European markets. Azure and Databricks pipelines through to Power BI, generative AI in production, and models for risk, segmentation and fraud — with regulators on the other side of the table.
AzureDatabricksPower BISnowflakePythonGenerative AI -
Analytics Specialist (Lead) — RGBSI · client: Sony
Led a team of analysts, engineers and scientists driving data-informed decisions across the organisation, with governance, quality and security standards embedded in the team rather than bolted on afterwards.
Azure Data ServicesAWSAirflowSnowflakePower BI -
BI Reporting Lead — TheMathCompany · clients: Hudson's Bay, Sobeys
Analytics strategy and capability roadmaps for large retailers. Built and mentored delivery teams, and set the quality, lifecycle and engineering standards used across engagements.
AzureAWSAirflowSnowflakePython -
BI Reporting — Mindtree · client: Microsoft
Browser log and Bing Ads analytics. Owned data acquisition and delivery for operational, client and KPI reporting, and designed distributed pipelines processing data at scale and in real time.
SQLOracleDB2MicroStrategyAzure Synapse -
Risk Reporting & Analytics — Infostretch · client: Wells Fargo
Risk scorecards built with functional partners, root-cause analysis on performance drivers, and sensitivity, scenario and stress testing across complex financial services data.
SQLETLScenario modellingScorecards -
Business Analysis & Reporting — Fidelity Investments
Market and transaction data collected through OLTP systems and REST services, cleaned and normalised, then used to train time-series and random forest baselines back-tested against production history.
TableauSQLPythonSparkOracleDB2 -
Retail & Wholesale Banking — iGATE · clients: Royal Bank of Canada, Bank of Tokyo-Mitsubishi
Estimation, project management and stakeholder coordination, authoring BRDs and FSDs. QA across traceability, defect tracking and status reporting, and ETL and warehouse delivery with upstream load monitoring.
SQLOracleDB2ETL testingBRD / FSD
Education and qualifications
CFA Program — Level III candidate
CFA Institute. Relevant here mainly because this work sits where finance, risk and data meet, and it helps to speak all three without a translator in the room.
B.Tech, Electrical & Electronics Engineering
SASTRA University, 2007–2011.
Reporting & BI
Modelling & context
Product analytics
AI & retrieval
Evals & observability
Platform
Languages
What I have been writing about
How every analytics role changed in eighteen months
The headline everybody expected was replacement. What happened instead was a shift in which part of each job is scarce. In almost every analytics role, the…
ReadYour semantic layer is your context layer
Every company that has tried to put an assistant in front of its data has run the same experiment, whether or not they meant to. Point a capable model at the…
ReadText-to-SQL fails on your warehouse, and it is not the model's fault
Text-to-SQL benchmarks look encouraging. Then somebody runs the same model against a real warehouse and accuracy falls off a cliff, and the conclusion drawn is that…
ReadWhy nobody opens your Power BI reports
A company spends nine months building a reporting layer. Usage reports show forty opens in the first week and six a month later. The conclusion that gets drawn is…
ReadGetting a Power BI semantic model ready for Copilot
The pattern is consistent enough that we can usually predict it before opening the file. Copilot gets switched on. There is a burst of enthusiasm. Within about six…
ReadEvals are the new unit tests
Ask a team how they know their AI feature works and you get one of three answers. "It seems good." "We tried a few questions." Or, occasionally, a number — and that…
ReadTopics I am happy to talk about
- Metric definition and semantic modelling in dbt and Power BI
- Putting generative AI into a regulated decision path
- Why analytics roles changed in eighteen months, and what to hire for now
- Building an eval suite before you build the assistant
- Autonomous KPI monitoring that people do not mute
Available for advisory work, technical due diligence on an AI or analytics estate, and the occasional talk or podcast. Email is the fastest route: hello@proanalytiqs.com.
Happy to talk, even if it goes nowhere.
If you are weighing up an AI project and want an honest read on whether your data can carry it, that is a conversation worth having whether or not you ever work with us.
Usually a reply the same working day · IST (UTC+5:30)