IDP/OCR Engineer

Shakti Solutions; United States; On-site; USD 90,000 - 150,000; Full time

Job summary

Shakti Solutions in the United States is seeking a dedicated IDP / OCR Engineer to design and operate production document processing pipelines. You'll work with OCR tools like Tesseract, PaddleOCR, and cloud services, handling diverse inputs from photos to scanned PDFs, emphasizing accuracy and robust testing.

You will build and maintain extraction across multiple document types, implement image quality gates, and collaborate with forensics to calibrate tamper scores, while creating regression

Qualifications

• BS or MS in computer science, engineering, or related field or equivalent practical experience.
• 3+ years building or operating production document processing or OCR.
• Hands-on with OCR tools and image pre-processing.
• Python and queue-based or batch pipelines.
• Experience handling phone photos, scans, and PDFs.

Responsibilities

• Ensure documents are read accurately across 20 document types and inputs.
• Build image-quality gate: blur, glare, skew, resolution checks with outcomes.
• Calibrate tamper scores with forensics engineer to reduce false positives.
• Create regression tests that run on every change.
• Define field-level confidence scores reflecting error rates.
• Develop clear resubmission guidance for consumers.

Skills

3+ years document processing

Python

Testing & measurement

Messy inputs handling

Education

BS or MS in CS/engineering

Tools

Tesseract

PaddleOCR

Google DocAI

Image pre-processing

Job description

Full Time or Contract w2 with Shakti Solutions - No C2C or H1B transfer

Role Title: IDP / OCR Engineer

Required Skills

• BS or MS in computer science, engineering, or a related field, or equivalent practical experience.
• 3+ years building or operating production document processing or OCR.
• Hands‑on with OCR tools (such as Tesseract, PaddleOCR or a cloud OCR service such as Google DocAI ) and basic image pre‑processing (deskew, denoising, contrast).
• Python and experience with queue‑based or batch pipelines.
• Experience handling messy inputs: phone photos, scans and compressed PDFs.
• Disciplined about testing and measuring accuracy per document type.

Preferred Skills

• Layout‑aware or key‑value extraction models.
• Image‑quality assessment methods.
• Experience with eligibility, financial, insurance or identity documents.
• Government or regulated‑environment experience.

Responsibilities

To make sure documents are read accurately across all 20 document types and real‑world inputs. A key part of the job is telling “bad image” apart from “altered document,” so poor‑quality uploads are routed to resubmit rather than flagged as fraud. Maintain and improve extraction across 20 document types and their template variations.

Expected Deliverables

• Maintain and improve extraction across 20 document types and their template variations.
• Build the image‑quality gate: blur, glare, skew, resolution and compression checks, with clear outcomes (proceed, resubmit, route to review).
• Work with the forensics engineer to calibrate tamper scores by quality level, so poor‑quality documents don’t drive false positives.
• Build regression tests that run on every change.
• Define field‑level confidence scores that reflect actual error rates.
• Help design clear resubmit guidance for consumers.