Shivam Bhosale
VOL. 03 ED. 2026 TORONTO · ON STATUS · OPEN
Contact ↗
Data Analyst · AI Engineer · 3+ YRS BUILDING

Turning raw data
into decisions,
decisions into systems↗

I build across the full analytics stack — SQL & Python pipelines, Power BI dashboards, LLM agents, financial screeners, and locally-run AI systems. Currently a Data Analyst at MJR Capital shipping ML to production. Side hours go to agentic AI, reliability-first LLM pipelines, and a few too many side projects.

"Good decisions start with clean data."
SB / OPS REPORT · personal analytics LIVE
Experience
3+ YRS
Since Sep 2022 · grad school onward
Public Repos
124
▲ 18 last 90 days
Commit Activity · 12 wks
peak · Wk 47
Stack Mix
Python 42% · AI 28% · BI 18%
MODELS · shipped
CCY · CAD
UTC · —
SECTION · 01 / 04
Selected work.
Indexed · 5 projects · 2023–26
PRJ-01
DECISION ENGINEERING · FEATURED
Lodestar
AHPTOPSISMonte CarloPython
Personal decision-engineering system. Stacks AHP + TOPSIS + Monte Carlo simulation to weight, rank, and stress-test life and work decisions. The data-person's antidote to gut-feel — try the live engine below ↓
Methods
3
Status
WIP
Simulations
10k+
Live demo
↓
↓
PRJ-02
FINANCIAL DATA · CLI
Stonks.ca
PythonyfinancePandasRich
Real-time TSX equity & ETF screener for the terminal. Detects P/E anomalies, volume spikes, 52-week breakouts, and momentum — outputs plain-English signals and a CSV trade log. Built around a configurable signal-logic table.
Tickers / run
20+
Signal rules
10
API key
none
Export
CSV
↗
PRJ-03
GENAI · RELIABILITY
Honey Badger AI
GenAIOllamaStreamlitLocal
Multi-stage validation pipeline for LLM output. Every response is generated → critiqued → revised → rule-evaluated → confidence-scored → guardrailed before reaching the user. Fully local via Ollama, no API key.
Stages
6
Inference
local
Cost / query
$0.00
Guardrails
✓
↗
PRJ-04
LLM · MARKET INTEL
Bloomberg Replica
Claude APIYahoo FinanceECB FXPython
AI-powered market dashboard. Claude reads equities, FX, and macro from free APIs and writes an institutional-grade daily brief — for retail users, at zero subscription cost.
Subscription
$0
Sources
3+
Daily brief
auto
Tool-use
✓
↗
PRJ-05
DEEP LEARNING · CV
Agri-Optima
TensorFlowKerasDjangoCNN
Django + CNN platform for farmers: crop yield prediction, plant-leaf disease detection, soil analysis, and a live agriculture news feed. Built during graduate research with Agriculture & Agri-Food Canada.
Images trained
35k
Accuracy
95%
Δ vs baseline
+15%
Partner
AAFC
↗
LIVE · INTERACTIVE
Run a Monte Carlo. Right now.

A scaled-down Lodestar engine. Tune the three weights below, hit RUN, and watch 10,000 simulated decision trials draw a distribution in real time. This is what decision engineering actually looks like before the slide deck cleans it up.

IDLE · awaiting input
0.50
0.70
0.35
μ mean—
σ stdev—
p5—
p95—
trials0
verdict—
HISTOGRAM · decision score (0—100) n = 0
020406080100
SECTION · 02 / 04
Career timeline.
Range · 2022 — present
Role / Organization
2022
2023
2024
2025
2026
Data Analyst / AI EngineerMJR Capital · Mississauga
2025 — Present
AI Engineer · Grad ResearcherU. of Windsor × AAFC
2023 — 24
M.Applied ComputingU. of Windsor
2022 — 24 · MAC
Open-source · side projectsStonks · Honey Badger · Lodestar · 100+ repos
CONTINUOUS
2025 — Present Mississauga, ON
Data Analyst / AI Engineer · MJR Capital Services
Owning end-to-end financial & operational analytics. Python + SQL ETL on AWS, Power BI dashboards with DAX, and Scikit-learn models for classification, regression, and anomaly detection — wired back into dashboards as live KPIs. Built monitoring frameworks (precision, recall, RMSE, AUC) with stakeholder-facing reports.
Reporting time
−40%
ML models
PROD
Pipelines
ETL
Stack
AWS
2023 — 2024 Windsor, ON
AI Engineer · Grad Researcher · U. Windsor × AAFC
Federally partnered research with Agriculture & Agri-Food Canada. Trained CNNs on 35,000+ annotated images for plant disease detection, deployed real-time inference via a Django app with role-based access, and authored explainability reports for regulatory audiences.
Accuracy
95%
Δ baseline
+15%
Images
35k
Deploy
Django
SECTION · 03 / 04
Where the hours go.
Self-reported · past 12 months
TIME ALLOCATION · by domain n = 1 · self-reported
Analytics & SQL 32%
ML & Deep Learning 24%
LLMs / Agentic AI 20%
BI & Reporting 14%
Infra · Cloud 10%
A · 32%
Analytics & SQL
B · 24%
ML / Deep Learning
C · 20%
LLMs · Agentic AI
D · 14%
BI · Reporting
E · 10%
Cloud · MLOps
Analytics & SQL
A.01
  • Python — Pandas, NumPy
  • SQL · advanced joins, CTEs, window fns
  • PySpark
  • EDA · feature engineering
  • R · statistical modelling
ML & Deep Learning
B.02
  • Scikit-learn — classification, regression
  • TensorFlow · PyTorch · Keras
  • CNNs · anomaly detection
  • Hyperparameter tuning
  • Eval — precision, recall, AUC, RMSE
LLMs & Agentic AI
C.03
  • Claude API · OpenAI · Gemini
  • RAG pipelines · FAISS · LangChain
  • LangGraph · CrewAI
  • MCP · A2A protocols
  • Local inference · Ollama
BI & Reporting
D.04
  • Power BI · DAX
  • Executive dashboards
  • Power Automate
  • Tableau
  • Excel — pivot forensics
Databases
E.05
  • PostgreSQL · SQL Server
  • MySQL
  • MongoDB
  • Query optimization
  • Schema design
Cloud & MLOps
F.06
  • AWS — S3, EC2
  • Docker · CI/CD
  • Azure · GCP
  • Model lifecycle
  • Pipeline automation
AI App Dev
G.07
  • Django · FastAPI · Flask
  • Real-time inference endpoints
  • Streamlit
  • REST APIs
  • Role-based access
Currently exploring
H.08
  • Multi-agent orchestration
  • Long-context RAG · eval harnesses
  • dbt · modern data stack
  • Decision engineering — AHP, TOPSIS
  • LLM-as-judge
“
Good decisions start with
clean data.
Personal axiom
SB · 2026
SECTION · 04 / 04
Get in touch.
Reply time · same day

Hiring an analyst who
thinks like an engineer?
Let's talk.

CH-01 · Email
shivambhosale909
@gmail.com
primary · always read
CH-02 · LinkedIn
/in/
shivambhosale
recruiter DMs open
CH-03 · GitHub
@ShivamBhosale
124 public repos
CH-04 · Location
Toronto
Ontario, CA
hybrid · remote