Principal engineer / AI, data, and analytics systems

Akshay Sardana

I build production AI and data systems for complex business workflows.

Most of my work sits between ambiguous business questions and reliable technical systems: conversational analytics, anomaly investigation, commerce automation, privacy-aware data infrastructure, and large-scale data platforms.

Data scale 10TB+ daily platform workloads
Platform reach Shared standards across multiple business units
Delivery 100+ pipelines and data models in 6 months
Analytics lift 60% reduction in analytics backlog

Work

Practical systems for AI, analytics, and data-heavy operations

The common thread is reliability: taking work that is ambiguous, manual, or fragile and turning it into systems with clearer data, behavior, and ownership.

Applied AI and Analytics Systems

Production AI and analytical systems that turn ambiguous business, editorial, commerce, and operational signals into governed, reviewable decisions.

  • Conversational analytics, anomaly investigation, classification, and defect-detection workflows
  • LLM workflows with structured outputs, guardrails, review gates, and observable failure modes
  • Retrieval, classification, multimodal verification, and privacy-aware NLP patterns

Data Platform Strategy

Practical architecture and sequencing for fragmented pipelines, warehouse models, and ownership boundaries.

  • Architecture and sequencing across BigQuery, dbt, Airflow, streaming, and PySpark
  • Data quality, lineage, and model contracts for shared datasets
  • Operating model for roadmap, support, and stakeholder intake

LLM Workflow Reliability

Auditable AI workflows with deterministic checks, structured outputs, tool use, human review, and production observability.

  • Guardrails for confidence, safety, retries, and failure investigation
  • Evaluation harnesses for SQL generation, anomaly narratives, page verification, and classification
  • Review gates for multimodal, classification, agentic, and privacy-sensitive systems

Analytics Engineering and Self-Service

Semantic models, dashboard contracts, and enablement practices that let analysts and business teams answer repeat questions with less ad hoc support.

  • Reusable dbt models and metrics with tested definitions
  • Self-service paths that keep sensitive logic governed
  • Training and documentation for engineers, analysts, and operators

Selected Systems

A few representative areas

These are short versions for context. The fuller technical notes keep confidential details omitted but go deeper on constraints, architecture, and outcomes.

AI Analytics

Conversational Analytics Agent

A governed natural-language analytics agent that turns business questions into validated BigQuery analysis across Ads, Editorial, Commerce, and operations workflows.

  • AI analytics
  • conversational BI
  • SQL safety
  • semantic retrieval
Applied AI Systems

AI Anomaly Analysis and Classification Systems

Applied AI and statistical analysis workflows that classify content, investigate performance anomalies, and surface operational changes with structured outputs, rules, and reviewable evidence.

  • anomaly analysis
  • LLM classification
  • statistical detection
  • business rules
Data Platform

Shared Data Platform and Self-Service Analytics

A shared analytics foundation for high-volume digital media and commerce data, built to reduce repeated requests and increase safe self-service.

  • data platform
  • semantic layer
  • self-service analytics
  • data quality
AI Workflow Automation

AI Commerce Defect Detection

A reviewable AI workflow that detects commerce catalog and retailer-page defects before operational issues compound.

  • multimodal review
  • catalog validation
  • operator workflow
  • confidence scoring

Approach

Start with the workflow, then make the system reliable

The pattern is direct: understand the decision or user path, ground it in trustworthy data, design for production behavior, and leave teams with maintainable operating practices.

Clarify

Start with the business workflow

Define the decision, user path, source systems, owners, failure modes, and practical constraints before choosing the AI or data architecture.

Ground

Connect AI to governed data

Use source-of-truth datasets, metric contracts, retrieval metadata, and access boundaries so answers can be traced and corrected.

Operate

Design for evals and production behavior

Build dry runs, tests, reconciliation checks, confidence gates, observability, and human review into the workflow instead of adding them after launch.

Transfer

Leave durable systems and team capability

Document decisions, coach internal owners, and leave an operating model that can keep improving after the initial build.

Experience

Hands-on principal engineering across AI, data, and platform work

I usually work where strategy, architecture, implementation, and operating ownership need to meet.

Principal Engineer, AI/ML/Data - Hearst Magazines

Principal Engineer since 2023, promoted from Senior Data Engineer, working inside a large digital media and e-commerce publisher with 140M+ monthly consumers.

  • Lead cross-functional data and AI work across roadmap, architecture, stakeholder alignment, and delivery
  • Platform impact across multiple engineering, analytics, and business teams
  • Build conversational analytics, anomaly analysis, commerce validation, article classification, and shared data-platform systems

Senior Data Engineering - Meta, Point Predictive, 1stDibs, Barclays

Prior work across consumer-scale event pipelines, fintech model infrastructure, e-commerce ML/data systems, and financial technology.

  • Worked on real-time NLP PII detection, 5B+ daily event streams, and privacy-aware analytics at Meta
  • Improved fintech and e-commerce ML infrastructure through derived data, feature engineering, SageMaker migration, and AWS ML services
  • Built across dbt, Airflow, PySpark, Kafka/Kinesis, BigQuery, AWS, GCP, Azure, LangGraph, OpenAI, Anthropic, Bedrock, and Vertex AI

Systems leadership pattern

Hands-on principal engineering work where strategy, architecture, implementation, and operating ownership need to converge.

  • Translate ambiguous mandates into scoped delivery plans and measurable system behavior
  • Build enough of the critical path to prove the architecture under real constraints
  • Create contracts, documentation, and team practices that survive beyond the initial launch

Contact

Continue the conversation

Email or LinkedIn are usually best for notes, context, or continuing a conversation. Calendly is there when a scheduled chat is easier.