Saif Mohammed

• Owned end-to-end design and delivery of applied ML solutions — from problem scoping
with clients to production deployment — operating with CTO-level ownership across the
full product lifecycle.
• Built and shipped a production ML-driven influencer-brand matching system, applying
feature engineering and supervised learning on audience signal data to deliver measur
able client outcomes.
• Engineered cross-platform AI workflows integrating Gemini API, Meta Graph API, Microsoft
Entra ID, and TikTok API, enabling intelligent automation across enterprise data systems.
• Collaborated directly with business stakeholders to understand pain points, translate
them into technical requirements, and rapidly iterate on LLM-powered solutions — closing
the last-mile gap in AI adoption.
• Built a smart task-notification system using Microsoft identity infrastructure, demonstrat
ing the ability to work across engineering and organizational tooling in a cross-functional
environment.

• Contributed to several Open-source repositories: Kestra-io, OpenLibrary, Dodo Payments
and more.
• Fixed Bugs, Backend Flows and worked on front-end issues.

• Designed and shipped a Python production system end-to-end — an automated purchase
order pipeline processing 500+ orders with 99.8% accuracy, replacing 4 hours of daily
manual work.
• Applied ML-based pattern matching and feature engineering to build an intelligent file
segregation engine, reducing processing time by 97% across 15+ sellers.
• Built production-grade error handling, logging, and monitoring infrastructure maintain
ing 99.9% system uptime — demonstrating applied ML engineering rigor in live deploy
ment.
• Prototyped and deployed AI-assisted creative automation workflows using n8n, increas
ing social media engagement by 130% through data-driven content scheduling.

• Built and deployed a video recommendation model using content-based filtering (cosine
similarity), increasing CTR by 20% on an EdTech platform.
• Owned the full ML pipeline: data preprocessing (Pandas/NumPy), feature extraction
(CountVectorizer), model training, evaluation, serialization (pickle), and REST API deploy
ment (Flask).

• 8+ merged PRs to Kestra (workflow orchestration for ML pipelines) — debugged complex
data propagation bugs in branching execution flows and improved exception handling
across distributed pipeline runs.
• Contributed to Apache Airflow security: hardened credential masking in DAG execution
logs, improving data governance for ML pipeline deployments.
• Demonstrated ability to quickly read, understand, and fix issues in large, unfamiliar code
bases — a core requirement for applied ML work across diverse enterprise systems.
• Worked across engineering and product surface areas (billing, UI, orchestration) — reflect
ing the cross-discipline collaboration the role requires.

• Debugged and resolved a data propagation issue in Flowable tasks (branching logic),
ensuring input variables remain accessible during complex execution flows.
• Worked on enhancing exception handling flows, improving error diagnostics, masking
sensitive data and adding new features.
• Authored the CSV export functionality for the Usage Table component, enabling users
to extract and analyze complex billing data (usageHistory, apiCost) externally, improving
data portability and reporting capabilities.
• Developed and optimized core billing UI components (including Pricing Tables and
Payment Cards), ensuring type-safe integration and standardized error diagnostics for
downstream developers.
Rootz
• Built an internal tool tracking client interations and progress on the opportunities
• Engineered a machine learning algorithm on an influencer marketing platform to match
brands with their ideal audience.
• made a smart notification system for assigning and completion of tasks within the organ
isation using microsoft entra id and microsoft api
• Used n8n workflows to optimize advertisment campaigns across different platforms
including Google, Meta and more.
• Worked with abound APIs including Meta Graph API, Microsoft API, Gemini API, TikTok API
and more for optimization of ad campaigns using Artificial Intelligence.

• Built an automation software for inventory management which processed incoming pur
chase orders.
• Engineered an NLP pipeline using Logistic Regression to automate the classification of
high-volume procurement emails, improving inventory processing efficiency
• Streamlined inventory management and report generation process.

Multi-agent RAG system for automated scientific literature discovery, reasoning, and synthe
sis.
• Designed and implemented a multi-agent sequential reasoning pipeline using Google
ADK and Gemini 2.0 Flash for large-scale NLP inference a RAG pipeline integrating arXiv API a source, enabling grounded,
citation-backed research report generation end-to-end via Flask, the full lifecycle from retrieval and reasoning to
structured PDF output

• Developed Automated Purchase Order System using Python, openpyxl, and IMAP/SMTP
protocols to eliminate manual segregation of Amazon purchase orders, reducing process
ing time by 90%
• Implemented Intelligent file segregation engine that automatically distributes products
to 15+ sellers based on barcode matching, processing 500+ line items in under 2 minutes
vs 4 hours manually
• Constructed Robust Email Automation Pipeline with real-time monitoring, command
parsing, and template-based Excel generation, handling 500+ purchase orders with 99.8%
accuracy.
• Engineered Production-Ready Error Handling with comprehensive logging, duplicate de
tection, and automated email notifications maintaining 99.9% uptime in live deployment
• Visualized Data and formulated performance insights using AgencyAnalytics, Datorama
and PowerBI
• Used Automation Tool n8n to automate creative posts on social media platforms, bringing
engagement up by 130%