Skip to the work

AI PRODUCT BUILDER · MACHINE LEARNING ENGINEER

Mohamed El Askary

AI product builder and machine learning engineer. I take systems from zero to one: ingestion to model to production to the customer.

Founder & CEO of Vantage Systems. Senior Data Scientist. United States Air Force veteran with an active U.S. security clearance. Delivery across the U.S., MENA, Europe, and South Asia.

3 REGIONS

live enterprise deployments

50+ KPIS

computed by the Vantage metrics engine

0 → 1

how every project here was built

View the workGet in touch
Portrait of Mohamed El Askary

EL ASKARY, MOHAMED · IRVINE, CA

02 / SELECTED WORK

Selected Work

Machine learning first, then the systems it runs inside. Open any card for the full build. Every number comes from a committed artifact, an eval harness or a training report, and every limitation those artifacts show is quoted with it.

GROUP A

Machine Learning

Sensor prognostics, industrial vision, data-centric ML, applied-AI decisioning and clinical risk. Public work carries its real artifacts; client work carries an architecture card.

01PUBLICPREDICTIVE MAINTENANCE

12.95RUL RMSE · beats every single model

Horizontal bar chart of per-sensor Pearson correlation with clipped remaining useful life on FD001 training data. Sensors 11, 4 and 15 show the strongest negative correlations near -0.75; sensors 12, 7, 21 and 20 the strongest positive correlations near +0.7.
FD001 · per-sensor correlation with clipped RUL

02PUBLICINDUSTRIAL COMPUTER VISION

67.1%mAP@50 across six defect classes

Roboflow workflow canvas showing the inference pipeline: input image into the object detection model, into a detections filter block, into an email notification block.
Managed alert path · model → detections filter → notification

03PUBLICDATA-CENTRIC ML

100%PII recall · zero leaks in the export

Terminal output of the evaluate command showing the five release gates with required and actual values: two gates marked FAIL and three marked pass, followed by a release blocked banner explaining the dataset does not ship in this state.
Release gates · two failing, left failing

04PUBLICFRAUD & AML DECISIONING

0 / 471false-positive declines, 471 legitimate txns

Terminal table of eval results listing fraud typologies detected 5 of 5, fraud transactions not auto-approved 26 of 32, false-positive declines 0 of 471, AML routed to human 7 of 7, and cost of the run $0.7612 against a $1.00 target.
Eval harness · scored against seeded ground truth

05PUBLICBRAND THREAT DETECTION

300 → 0candidates checked to pages sent

Terminal output listing ten registered lookalike domains with their permutation technique, registration age, triage classification and severity score, all held in the digest below the paging threshold.
Live findings · ten typosquats, all held below the page threshold

06DELIVEREDHEALTHCARE RISK MODELLING

0.8533ROC AUC on four significant predictors

ROC curve for the reduced four-feature logistic regression model, bowing well above the random-classifier diagonal, with area under the curve 0.8533.
ROC curve · reduced model, AUC 0.8533

07DEPLOYED · PRIVATEINDUSTRIAL SENSOR TELEMETRY

3regions on condition-based maintenance

08DEPLOYED · PRIVATEGEOSPATIAL ML

3data families fused into one forecast

09DELIVEREDSPACE SYSTEMS

0 → 1environmental data to mission parameters

GROUP B

Production Systems

The platforms, contracts and pipelines behind the models, built and operated end to end.

01DEPLOYED · PRIVATEOPERATIONAL INTELLIGENCE

50+live KPIs, with four agents on top

02PUBLICAGENT TOOLING & EVALS

5 / 5seeded blockers caught, zero false positives

GitHub pull request page showing the agent's structured review comment: a request-changes verdict, a summary, and a finding citing rule SEC-001 for a SQL injection introduced by string interpolation.
Live review on PR #1 · seeded SQL injection, caught in ~20 s

03DEPLOYED · PRIVATEREGULATED LENDING

1platform replacing a stack of tools

04DEPLOYED · PRIVATEDEBT SERVICES

2delivery phases, both shipped

03 / VENTURES

Ventures

The three companies where I own the outcome, not just the code.

OPERATIONAL INTELLIGENCEMULTI-TENANT SAASLIVE · 3 REGIONS

Vantage Systems

Founder & CEO

An enterprise operational intelligence platform that consolidates financial, production and supply chain data from client source systems into a unified executive reporting layer. Live with multi-entity clients across the U.S., Pakistan and the Gulf.

  • I architected and built the platform personally: schema-per-tenant PostgreSQL, Kafka and SQS event streaming, Redis caching, an S3 and GCS data lake, six-role RBAC governing subsidiary versus group-wide visibility, and the Next.js front end on top.
  • I built a metrics intelligence engine computing 50+ KPIs with predictive forecasting, threshold alerting and cross-entity benchmarking, plus four specialised analytics agents running audit trails, real-time anomaly detection and workflow routing.
  • I own the complete lifecycle: PRDs and technical specifications, sprint planning, QA, deployment, post-launch telemetry, go-to-market, pricing and contract negotiation.
  • I run the enterprise sales cycle end to end (discovery, tailored technical proposals, security and procurement review, pricing, negotiation), and I have closed with publicly traded industrial groups.
getvantage.global

AI VOICE & WORKFLOW INFRASTRUCTURE · MENA EXPANSION

MosarcAI

Co-Founder & CEO

A U.S.-based AI automation and systems company building voice and workflow infrastructure for organisations with recurring operational load. Its systems answer inbound calls, gather the first round of information, verify caller intent and start structured workflows, holding continuity as work moves between departments.

  • The systems integrate with the CRMs, calendars, messaging tools and task platforms a business already runs rather than displacing them, so a team adapts gradually instead of absorbing a migration.
  • I lead the venture's strategic expansion into the Middle East, announced in December 2025 after accelerating U.S. adoption and sustained inbound interest from Gulf-region organisations.
  • That expansion meant fitting the infrastructure to the region rather than shipping the U.S. build: multilingual and dialect-compatible voice handling, and workflow structures that match regional communication patterns.
  • The systems introduced into the region are aimed at large customer-facing teams that depend on predictable communication handling: healthcare clinics, real estate groups, hospitality networks, automotive service providers and professional service operations.
  • The expansion strategy is long-term operational partnership rather than transactional sale: measurable pilot periods, structured adaptation and continuing evaluation of workflow patterns.

ADVISORY · MULTI-INDUSTRY

SHEM Consulting Inc.

Technology, AI & Business Consultant

I advise client organisations on where AI genuinely creates operational value and where it does not, then build the thing rather than leaving a recommendation behind. Engagements run across multiple industries and have since 2022.

  • AI value assessment. I work out which problems in a business are worth automating and which are not, and I say so plainly. Telling a client that their best opportunity is a process fix rather than a model is part of the job, not a failure of it.
  • Stack and data-infrastructure evaluation. I assess the systems a client already runs, where the data actually lives, and what has to be true before anything intelligent can be built on top of it.
  • Automation and internal tooling. I design and build the tooling and AI-assisted workflows the assessment calls for, and I stand them up inside the team that will operate them.
  • Market research and process analysis. I run stakeholder interviews and current-state assessments to surface process gaps, then convert the findings into prioritised, actionable operational recommendations.
  • Training. A large part of every engagement is teaching non-technical staff to use these tools well, because a workflow nobody understands gets abandoned the first week it is inconvenient.
  • Implementation through to adoption. I present directly to client leadership, translate quantitative findings into decision-ready guidance, then support the work into production and track outcomes against the stated objectives, adjusting on measured results rather than on the original assumptions.

04 / EXPERIENCE

Experience

I have built inside enterprise technology, medical device, space, construction and the military. Here is what I owned in each, in the order it matters.

  1. CEO & Founder · Vantage Systems

    • I founded and lead an enterprise operational intelligence platform, deployed with multi-entity clients across the U.S., Pakistan and the Gulf.
    • I architected the multi-tenant backend personally and built the 50+ KPI metrics engine and the four production analytics agents that sit on top of it.
    • I own the complete product lifecycle and the enterprise sales cycle, closing with publicly traded industrial groups.

  2. Senior Data Scientist · GeoAct

    • I build predictive maintenance models on industrial sensor telemetry, forecasting remaining useful life and failure probability so clients can move from fixed-schedule to condition-based intervention.
    • I build natural hazard and disaster risk models that fuse multi-source earth observation, remote sensing and ground data.
    • I own the full pipeline (ingestion, feature engineering, training, validation, deployment against live data), and I design the detection and alerting logic that balances sensitivity against false positives.

  3. AI Builder · RedLantern Studios (By Red, LLC)

    • I build across the studio product portfolio in TypeScript and Next.js alongside the founder and CEO, spec-first, with AI-assisted development as the primary method.
    • I contribute to Authentic Hadith, QuietBuild-OS and the QBos founder platform, Rushd and RobbyPA.
    • I own growth and partnership work, measured through studio analytics against a defined attribution method.

  4. Product Builder, Financial Services · Independent Client Engagements

    • I built and deployed a multi-tenant CRM with an integrated power dialer for a reverse mortgage lender, in a regulated lending context.
    • I built a lead generation and routing platform for a debt relief company with certified consent capture, telephony routing and transactional email, delivered in two phases.

  5. Data Analyst · Edwards Lifesciences

    • I rebuilt the data pipelines and ETL workflows serving cross-functional reporting, cutting report generation time 40% while improving accuracy.
    • I traced recurring data quality gaps to root cause and shipped automated validation at ingestion, cutting manual correction hours 25%.
    • I built predictive models and executive dashboards tracking 15+ KPIs across business units, informing quarterly strategic planning.

  6. Data Analyst · Johnson & Johnson MedTech

    • I standardised data ingestion across clinical trial datasets with incompatible structures, cutting preparation time for cross-functional analysis 30%.
    • I developed predictive models and statistical analyses supporting device product strategy, roadmap prioritisation and launch sequencing with engineering and clinical teams.
    • I produced the analysis informing executive go / no-go decisions on product development investment.

  7. Consultant · SHEM Consulting Inc.

    • I deliver market research and business process analysis across industries, converted into prioritised operational recommendations.
    • I present analysis directly to client leadership, translating quantitative findings into decision-ready guidance for non-technical audiences.
    • I track post-implementation outcomes against stated objectives and adjust on measured results rather than on initial assumptions.

  8. Project Engineer · Layton Construction

    • I managed schedules, budgets and contractor coordination on active construction sites, holding milestone delivery through daily progress tracking.
    • I interpreted technical blueprints, prepared cost estimates and maintained change order documentation across multi-stakeholder workflows.

  9. Data Analyst · Egyptian Space Agency

    • I built geospatial predictive models supporting satellite mission planning and remote sensing analysis, processing large-scale environmental datasets to inform mission parameters.
    • I delivered analytical reports and visualisations to senior mission leadership alongside a multidisciplinary team of engineers and scientists.
    • I worked across institutional boundaries in an Arabic-language technical environment, coordinating with international partners on shared data standards.

  10. Research Assistant · Zewail City of Science and Technology

    • I ran experiments and data analysis in electrical engineering research under senior faculty.
    • I supported experimental design, data collection and statistical analysis across multiple workstreams, contributing to publication-ready deliverables.

  11. Airman & Squadron Commander · United States Air Force

    • I led 40 airmen as Squadron Commander in tactical operations, land navigation exercises and mission-critical team coordination.
    • I completed foundational military training, leadership development and academic coursework at the United States Air Force Academy.

05 / TECHNICAL TOOLKIT

Technical Toolkit

I split my toolkit by the half of the problem I am standing in: the thing I am building, the model inside it, the data feeding it, and the business case that has to survive the room.

P-01

Engineering & Product

  • TypeScript / JavaScript
  • Next.js
  • Python
  • SQL
  • PostgreSQL
  • Kafka
  • AWS (S3, SQS, Lambda)
  • Redis
  • GCS
  • REST APIs
  • Multi-tenant architecture
  • RBAC
  • CI/CD
  • Technical specs & PRDs
P-02

AI & Machine Learning

  • Spec-driven AI-assisted development
  • LLM orchestration
  • Multi-agent systems
  • Predictive modelling & forecasting
  • XGBoost
  • LSTM
  • Anomaly detection
  • Remote sensing & geospatial ML
  • Multi-source data fusion
P-03

Data & Analytics

  • ETL pipelines
  • Data quality & validation
  • Root cause analysis
  • KPI framework design
  • Statistical analysis
  • Executive dashboarding
  • Tableau
  • Alteryx
  • R
P-04

Finance & Business

  • Financial modelling
  • GL & multi-entity consolidation
  • Forecasting & budgeting
  • Pricing strategy
  • Contract negotiation
  • Enterprise sales

Languages: English (native), Arabic (native), French (proficient).

06 / MEDIA & RECOGNITION

Media & Recognition

Business Insider / Newsfile

MoSarcAI Strategic Expansion Into the Middle East

newsfilecorp.com · syndicated to Business Insider Markets

The SCIAC

SCIAC Friday Spotlight: Mohamed El-Askary, Chapman Swim & Dive

thesciac.org

07 / EDUCATION & SERVICE

Education & Service

Active U.S. Security Clearance

08 / CONTACT

Contact

If you're building something that matters, I answer my email.

MOHAMED EL ASKARY · IRVINE, CA

OPEN TO WORK & CONSULTING ENGAGEMENTS