Ashwathy Ashokan

Data Analyst | Senior Statistical Programmer

Senior data professional with 5+ years building enterprise-scale analytics pipelines, ML systems, and clinical research solutions. Specializing in SAS, Python, SQL, PySpark, and cloud infrastructure.

Portrait of Ashwathy Ashokan

About Me

I'm a passionate data professional with a unique background bridging clinical research and modern data science. My journey began in pharmaceutical data management at Agati Clinical Informatics, where I mastered CDISC standards (SDTM/ADaM) and led FDA regulatory submissions for Phase I-III clinical trials across multiple therapeutic areas including Oncology, Endocrinology, and Cardiovascular.

Over the past 5+ years, I've evolved into a full-stack data professional, combining statistical rigor with machine learning and big data technologies. I excel at building production-grade data pipelines—from real-time Kafka streaming to predictive modeling with XGBoost—while maintaining the highest standards of data quality, reproducibility, and compliance.

I'm driven by complex problems that require technical depth and strategic thinking. Whether automating clinical workflows, scaling data pipelines to handle millions of events daily, or delivering ML models to production with AWS and Docker, I approach every challenge with precision, innovation, and a focus on measurable impact.

Core Strengths:

  • 5+ years statistical programming & CDISC expertise
  • End-to-end ML & predictive modeling systems
  • Big data pipelines: Kafka, Spark Streaming, Hadoop
  • Real-time data visualization & interactive dashboards
  • Cloud infrastructure & DevOps (AWS, Docker, Airflow)

Quick Contact

Location

Ontario, Canada

Skills & Expertise

Comprehensive toolkit for modern data challenges

💻

Programming & Data

SAS Base/Advanced/Macros (5+ years)PythonSQLPySparkRScala
🤖

ML & Analytics

XGBoostScikit-LearnTensorFlowNLPTopic ModelingSentiment AnalysisRegressionClustering
☁️

Big Data & Cloud

Apache KafkaSpark StreamingHadoopAWS EC2DockerApache Airflow
📊

BI & Visualization

TableauPower BIPlotlyMatplotlibSeabornD3.js
⚕️

Clinical & Regulatory

SDTM/ADaMDefine.xmlP21 SubmissionsTLFsFDA ComplianceCDISC21 CFR Part 11
🗄️

Databases & Tools

MongoDBPostgreSQLFastAPIGitJupyterTwilioNext.js

Experience

5+ years building enterprise-scale data systems

Senior Statistical Programmer

Agati Clinical Informatics

Ontario, Canada

2019 — 2025

Led end-to-end data programming and analytics initiatives across 8+ Phase I-III clinical trials in Oncology, Endocrinology, and Cardiovascular therapeutic areas.

  • ✓Designed and validated SDTM/ADaM datasets ensuring FDA submission readiness; implemented automated QC workflows achieving 99.9% data accuracy
  • ✓Developed reusable SAS macro library with 30+ enterprise-grade macros, reducing programming time by 40% across teams
  • ✓Engineered Define.xml and Pinnacle 21 submission packages ensuring 21 CFR Part 11 compliance for regulated submissions
  • ✓Optimized TLF generation pipeline from 3 hours to 15 minutes through automation and code optimization
  • ✓Mentored 5+ junior statistical programmers; created standardized templates and SOPs adopted company-wide
  • ✓Collaborated with Biostatistics, Data Management, and global study teams across multiple time zones

Data Analyst & ML Engineer

Personal & Academic Projects

Remote

2020 — Present

Architected and deployed production-grade machine learning and big data solutions for real-world applications.

  • ✓Built end-to-end NLP streaming pipeline (iBrand) processing 10K+ news articles daily, achieving 92% sentiment classification accuracy
  • ✓Engineered Canadian housing price prediction system with XGBoost achieving $45K RMSE; deployed weekly retraining on AWS EC2 with Airflow orchestration
  • ✓Designed Kafka-based real-time streaming infrastructure handling 1M+ events/day with sub-100ms latency guarantees
  • ✓Developed Next.js interactive dashboards delivering real-time business intelligence to non-technical stakeholders
  • ✓Integrated Twilio SMS alerts for time-sensitive predictions; automated workflows reduced manual tasks by 85%
  • ✓Containerized microservices with Docker; implemented Git-based CI/CD for reproducible, auditable deployments

Featured Projects

End-to-end systems solving real-world challenges

iBrand – AI/NLP Big-Data Pipeline

Big Data & ML

Streaming NLP platform for real-time brand intelligence and automated social content generation.

Problem

Manual analysis of news articles for brand mentions was time-consuming and inconsistent.

Solution

Built an end-to-end streaming pipeline: multi-source news scrapers → Kafka → Spark Streaming NLP (cleaning, NER, sentiment, topic modeling) → MongoDB → FastAPI backend.

Impact

  • ▸Processes 10,000+ articles daily with 92% sentiment accuracy
  • ▸Reduces manual analysis time by 95%
  • ▸Enables automated, brand-tone social media posts
  • ▸Containerized microservices for elastic scaling

Tech Stack

PySparkKafkaSpark StreamingMongoDBFastAPIDockerTransformers
View on GitHub→

Canadian Housing Price Prediction

ML & Analytics

Automated ML system predicting residential property values with real-time alerts and dashboards.

Problem

Real estate investors needed faster, data-driven property valuation insights.

Solution

Engineered automated pipeline: weekly Kijiji/Realtor.ca scraping → data cleaning & feature engineering → XGBoost training → Airflow orchestration on AWS EC2 → Twilio SMS alerts → Next.js dashboard.

Impact

  • ▸RMSE of $45K on validation set
  • ▸Weekly automated retraining and predictions
  • ▸SMS alerts for undervalued listings in real-time
  • ▸Interactive dashboard with feature importance visualization

Tech Stack

PythonXGBoostMongoDBAWS EC2AirflowTwilioNext.js
View on GitHub→

Clinical Data Automation & CDISC Pipeline

Data Engineering

Automated SDTM/ADaM generation system ensuring regulatory compliance and audit readiness.

Problem

Manual SDTM/ADaM creation and TLF generation consumed 3+ hours per analysis, creating quality risks.

Solution

Developed reusable SAS macro library with automated validation, Python scripts for Define.xml generation, and standardized specifications for 30+ pharmaceutical studies.

Impact

  • ▸TLF generation reduced from 3 hours to 15 minutes
  • ▸40% reduction in total programming time
  • ▸99.9% data accuracy achieved across submissions
  • ▸Reusable macro library adopted enterprise-wide for consistency

Tech Stack

SASPythonSQLDefine.xmlCDISCGitJupyter

Education & Leadership

Academic foundation and professional involvement

Big Data Analytics

Lambton College

2024–2025

Specialized curriculum in distributed computing, machine learning, data engineering, and big data technologies. Practical experience with Spark, Kafka, Hadoop, and cloud infrastructure.

BE Fire Technology & Safety

Noorul Islam University

2014–2018

Foundation in systematic problem-solving, risk assessment, and technical analysis. Developed analytical mindset and attention to detail applied to data science.

Professional Involvement & Leadership

Community

  • ▸PHUSE Member – Pharmaceutical User Software Exchange
  • ▸Symposium Coordinator – Conference organization & speaker liaison

Creative & Volunteer

  • ▸Magazine Editor – LaTeX & InDesign publication design
  • ▸Fire Safety Volunteer – Community education & consulting

Let's Connect

I'm always interested in hearing about new projects, opportunities, and collaborations.