Skip to content
BackNov 7
Juliano S.

Juliano S.

Agri-Commodity Data Scientist| Python Pipelines, AI Agents, Weather

I build data pipelines, models and AI agents for agriculture and commodity markets, and I have spent 14 years working inside those markets. Background: market research and trading-desk analytics in grains, oilseeds and fertilizers at Noble Group, COFCO International, Bloomberg LP, Aditya Birla Global Trading and a Dubai fertilizer tradIng Housa. Trained in quantitative agronomy and remote sensing. Today I run my own commodity research firm in Dubai, built on the same systems I offer here. On Upwork since 2017: 1,100+ hours, 100% Job Success. Recent work: An automated weekly research pipeline that pulls USDA, CONAB, CFTC, trade and port data and produces a 70-page grains and oilseeds deck without manual steps. An AI research assistant on WhatsApp that answers market questions from a curated database, with charts, sources and escalation to a human analyst. A futures backtest engine for corn and soybeans: 200 strategy variants tested with walk-forward holdouts, after finding and fixing three forms of look-ahead bias in the data. A historical WASDE archive and crop-condition panels from NOAA satellite data for the US, Brazil, Argentina and Ukraine. A crop-intelligence data service exposed to AI assistants through MCP. What I deliver: Market data pipelines: USDA (WASDE, export sales, crop progress), CONAB, CFTC COT, trade statistics, port lineups and weather, into clean, versioned datasets in Python, R, DuckDB and Parquet, with checks against source. AI agents and research automation: agents that query your data, produce charts with citations, draft market notes, and run in WhatsApp, Slack, Excel or a web app. Crop and weather analytics: yield models, Google Earth Engine and satellite vegetation indices, water-balance monitoring, extreme-weather tracking. Trading analytics: supply and demand balances, trade flows, freight and CFR parity, COT positioning, and backtests built to avoid look-ahead bias. Why it matters: anyone can build an agent now. The hard part is knowing when its output is wrong. I check numbers the way a trading desk does, before they reach you. How I work: clear scope, weekly delivery, documented code you own. English, Portuguese and Spanish.

🇦🇪Dubai, United Arab Emirates$100/hr100% JSS5.00 (11)Since Jun 2018

Earnings and projects

MRR
$8

Last 6 months ÷ 6

World#174,204 / 412k
🇦🇪Country#959 / 2,720
Revenue, last 30 days
$50

Last 6 months: $49

World#98,339 / 412k
🇦🇪Country#458 / 2,720
Total earnings
$32k

Billed in all: $32k

World#105,777 / 412k
🇦🇪Country#596 / 2,720
Per project
$2k

$49 in the last 6 months

World#76,998 / 412k
🇦🇪Country#472 / 2,720
Projects, last 6 months
1
100% fixed0% hourly
World#128,092 / 412k
🇦🇪Country#758 / 2,720
Projects in all
16

Across 22 contracts

59% fixed41% hourly
World#155,087 / 412k
🇦🇪Country#1,050 / 2,720

Earnings, last 30 days

$32.1k+$50 earned in the last 30 daysBilled in all, before Upwork's fee · drag across the chart to see what was earned in any stretch
$32,050$32,075$32,100$32,125Oct 8Oct 23Nov 7

Skill rankings: Top 32% in Time Series Analysis

Ranked by MRR among everyone on Upwork with the skill.

  1. #7Commodity Management19 freelancers · 🇦🇪 #1Top 37%
  2. #11ERDAS IMAGINE31 freelancers · 🇦🇪 #2Top 35%
  3. #64Agriculture & Forestry144 freelancers · 🇦🇪 #1Top 44%
  4. #81Time Series Analysis255 freelancers · 🇦🇪 #1Top 32%
  5. #113Environmental Science298 freelancers · 🇦🇪 #3Top 38%
  6. #118Google Earth246 freelancers · 🇦🇪 #2Top 48%
  7. #127Remote Sensing267 freelancers · 🇦🇪 #5Top 48%
  8. #284ArcGIS692 freelancers · 🇦🇪 #4Top 41%